[{"id": "proteingym", "features": [{"key": "epistasis.magnitude", "label": "Magnitude epistasis"}, {"key": "epistasis.sign", "label": "Sign epistasis"}, {"key": "epistasis.reciprocal_sign", "label": "Reciprocal sign epistasis"}, {"key": "diminishing_returns_index", "label": "Diminishing returns"}, {"key": "increasing_costs_index", "label": "Increasing costs"}, {"key": "global_idiosyncratic_index", "label": "Global idiosyncrasy"}, {"key": "r_s_ratio", "label": "Roughness-to-slope ratio"}, {"key": "gamma", "label": "Gamma"}, {"key": "gamma_star", "label": "Gamma star"}, {"key": "local_optima_ratio", "label": "Local optima ratio"}, {"key": "autocorrelation", "label": "Autocorrelation"}, {"key": "fdc", "label": "Fitness–distance correlation"}, {"key": "evolvability_enhancing_fraction", "label": "Evolvability-enhancing fraction"}, {"key": "global_optima_accessibility", "label": "Global optimum accessibility"}], "outcomes": [{"key": "site_independent", "label": "Site-Independent"}, {"key": "evmutation", "label": "EVmutation"}, {"key": "deepsequence_single", "label": "DeepSequence (single)"}, {"key": "deepsequence_ensemble", "label": "DeepSequence (ensemble)"}, {"key": "eve_single", "label": "EVE (single)"}, {"key": "eve_ensemble", "label": "EVE (ensemble)"}, {"key": "unirep", "label": "Unirep"}, {"key": "unirep_evotuned", "label": "Unirep evotuned"}, {"key": "msa_transformer_single", "label": "MSA Transformer (single)"}, {"key": "msa_transformer_ensemble", "label": "MSA Transformer (ensemble)"}, {"key": "esm_1b", "label": "ESM-1b"}, {"key": "esm_1v_single", "label": "ESM-1v (single)"}, {"key": "esm_1v_ensemble", "label": "ESM-1v (ensemble)"}, {"key": "esm2_8m", "label": "ESM2 (8M)"}, {"key": "esm2_35m", "label": "ESM2 (35M)"}, {"key": "esm2_150m", "label": "ESM2 (150M)"}, {"key": "esm2_650m", "label": "ESM2 (650M)"}, {"key": "esm2_3b", "label": "ESM2 (3B)"}, {"key": "esm2_15b", "label": "ESM2 (15B)"}, {"key": "wavenet", "label": "Wavenet"}, {"key": "rita_s", "label": "RITA S"}, {"key": "rita_m", "label": "RITA M"}, {"key": "rita_l", "label": "RITA L"}, {"key": "rita_xl", "label": "RITA XL"}, {"key": "progen2_s", "label": "Progen2 S"}, {"key": "progen2_m", "label": "Progen2 M"}, {"key": "progen2_base", "label": "Progen2 Base"}, {"key": "progen2_l", "label": "Progen2 L"}, {"key": "progen2_xl", "label": "Progen2 XL"}, {"key": "gemme", "label": "GEMME"}, {"key": "vespa", "label": "VESPA"}, {"key": "vespal", "label": "VESPAl"}, {"key": "vespag", "label": "VespaG"}, {"key": "protgpt2", "label": "ProtGPT2"}, {"key": "tranception_s_no_retrieval", "label": "Tranception S no retrieval"}, {"key": "tranception_m_no_retrieval", "label": "Tranception M no retrieval"}, {"key": "tranception_l_no_retrieval", "label": "Tranception L no retrieval"}, {"key": "tranception_s", "label": "Tranception S"}, {"key": "tranception_m", "label": "Tranception M"}, {"key": "tranception_l", "label": "Tranception L"}, {"key": "trancepteve_s", "label": "TranceptEVE S"}, {"key": "trancepteve_m", "label": "TranceptEVE M"}, {"key": "trancepteve_l", "label": "TranceptEVE L"}, {"key": "carp_38m", "label": "CARP (38M)"}, {"key": "carp_600k", "label": "CARP (600K)"}, {"key": "carp_640m", "label": "CARP (640M)"}, {"key": "carp_76m", "label": "CARP (76M)"}, {"key": "mif", "label": "MIF"}, {"key": "mif_st", "label": "MIF-ST"}, {"key": "esm_if1", "label": "ESM-IF1"}, {"key": "proteinmpnn", "label": "ProteinMPNN"}, {"key": "protssn_k_10_h_512", "label": "ProtSSN (k=10, h=512)"}, {"key": "protssn_k_10_h_768", "label": "ProtSSN (k=10, h=768)"}, {"key": "protssn_k_10_h_1280", "label": "ProtSSN (k=10, h=1280)"}, {"key": "protssn_k_20_h_512", "label": "ProtSSN (k=20, h=512)"}, {"key": "protssn_k_20_h_768", "label": "ProtSSN (k=20, h=768)"}, {"key": "protssn_k_20_h_1280", "label": "ProtSSN (k=20, h=1280)"}, {"key": "protssn_k_30_h_512", "label": "ProtSSN (k=30, h=512)"}, {"key": "protssn_k_30_h_768", "label": "ProtSSN (k=30, h=768)"}, {"key": "protssn_k_30_h_1280", "label": "ProtSSN (k=30, h=1280)"}, {"key": "protssn_ensemble", "label": "ProtSSN (ensemble)"}, {"key": "saprot_650m", "label": "SaProt (650M)"}, {"key": "saprot_35m", "label": "SaProt (35M)"}, {"key": "poet_200m", "label": "PoET (200M)"}, {"key": "mulan", "label": "MULAN"}, {"key": "prosst_k_20", "label": "ProSST (K=20)"}, {"key": "prosst_k_128", "label": "ProSST (K=128)"}, {"key": "prosst_k_512", "label": "ProSST (K=512)"}, {"key": "prosst_k_1024", "label": "ProSST (K=1024)"}, {"key": "prosst_k_2048", "label": "ProSST (K=2048)"}, {"key": "prosst_k_4096", "label": "ProSST (K=4096)"}, {"key": "escott", "label": "ESCOTT"}, {"key": "venusrem", "label": "VenusREM"}, {"key": "rsalor", "label": "RSALOR"}, {"key": "s2f", "label": "S2F"}, {"key": "s2f_msa", "label": "S2F-MSA"}, {"key": "s3f", "label": "S3F"}, {"key": "s3f_msa", "label": "S3F-MSA"}, {"key": "siterm", "label": "SiteRM"}, {"key": "esm3_open_1_4b", "label": "ESM3 open (1.4B)"}, {"key": "esm_c_300m", "label": "ESM-C (300M)"}, {"key": "esm_c_600m", "label": "ESM-C (600M)"}, {"key": "xtrimopglm_1b_mlm", "label": "xTrimoPGLM-1B-MLM"}, {"key": "xtrimopglm_3b_mlm", "label": "xTrimoPGLM-3B-MLM"}, {"key": "xtrimopglm_10b_mlm", "label": "xTrimoPGLM-10B-MLM"}, {"key": "xtrimopglm_1b_clm", "label": "xTrimoPGLM-1B-CLM"}, {"key": "xtrimopglm_3b_clm", "label": "xTrimoPGLM-3B-CLM"}, {"key": "xtrimopglm_7b_clm", "label": "xTrimoPGLM-7B-CLM"}, {"key": "xtrimopglm_100b_int4", "label": "xTrimoPGLM-100B-int4"}, {"key": "progen3_112m", "label": "Progen3 112m"}, {"key": "progen3_219m", "label": "Progen3 219m"}, {"key": "progen3_339m", "label": "Progen3 339m"}, {"key": "progen3_762m", "label": "Progen3 762m"}, {"key": "progen3_1b", "label": "Progen3 1B"}, {"key": "progen3_3b", "label": "Progen3 3B"}, {"key": "aido_protein_rag_16b", "label": "AIDO Protein-RAG (16B)"}, {"key": "protriever", "label": "Protriever"}], "records": [{"id": "A4_HUMAN_Seuma_2022", "label": "A4_HUMAN_Seuma_2022", "protein": "A4_HUMAN", "mean_mutations": 1.9462561609614475, "n_variants": 14811, "n_graph_configs": 14811, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5516256557487937, "epistasis.sign": 0.34250755332421423, "epistasis.reciprocal_sign": 0.10586679092699205, "diminishing_returns_index": -0.430586618171644, "increasing_costs_index": 0.2755444615740169, "global_idiosyncratic_index": 0.47455838225775143, "r_s_ratio": 0.713893930820098, "gamma": 0.6791699259576855, "gamma_star": 0.44576882609349167, "local_optima_ratio": 0.03234082776314901, "autocorrelation": 0.46376293922930506, "fdc": -0.2060303203789357, "evolvability_enhancing_fraction": 3.868871063423693e-05, "global_optima_accessibility": 0.9451083654040916}, "publication": {"title": "An atlas of amyloid aggregation: the impact of substitutions, insertions, deletions and truncations on amyloid beta fibril nucleation", "journal": "Nature Communications", "year": 2022, "doi": "10.1038/s41467-022-34742-3", "url": "https://doi.org/10.1038/s41467-022-34742-3"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.394, "evmutation": 0.385, "deepsequence_single": 0.449, "deepsequence_ensemble": 0.425, "eve_single": 0.312, "eve_ensemble": 0.316, "unirep": 0.345, "unirep_evotuned": 0.167, "msa_transformer_single": 0.337, "msa_transformer_ensemble": 0.351, "esm_1b": 0.302, "esm_1v_single": 0.316, "esm_1v_ensemble": 0.404, "esm2_8m": 0.382, "esm2_35m": 0.366, "esm2_150m": 0.372, "esm2_650m": 0.431, "esm2_3b": 0.48, "esm2_15b": 0.352, "wavenet": 0.261, "rita_s": 0.327, "rita_m": 0.281, "rita_l": 0.318, "rita_xl": 0.306, "progen2_s": 0.406, "progen2_m": 0.297, "progen2_base": 0.32, "progen2_l": 0.313, "progen2_xl": 0.314, "gemme": 0.478, "vespa": 0.251, "vespal": 0.186, "vespag": 0.487, "protgpt2": 0.487, "tranception_s_no_retrieval": 0.393, "tranception_m_no_retrieval": 0.273, "tranception_l_no_retrieval": 0.37, "tranception_s": 0.441, "tranception_m": 0.372, "tranception_l": 0.458, "trancepteve_s": 0.436, "trancepteve_m": 0.4, "trancepteve_l": 0.416, "carp_38m": 0.366, "carp_600k": 0.362, "carp_640m": 0.421, "carp_76m": 0.356, "mif": 0.279, "mif_st": 0.358, "esm_if1": -0.191, "proteinmpnn": 0.06, "protssn_k_10_h_512": 0.449, "protssn_k_10_h_768": 0.44, "protssn_k_10_h_1280": 0.417, "protssn_k_20_h_512": 0.41, "protssn_k_20_h_768": 0.406, "protssn_k_20_h_1280": 0.38, "protssn_k_30_h_512": 0.423, "protssn_k_30_h_768": 0.346, "protssn_k_30_h_1280": 0.372, "protssn_ensemble": 0.417, "saprot_650m": 0.411, "saprot_35m": 0.388, "poet_200m": 0.407, "mulan": 0.388, "prosst_k_20": 0.373, "prosst_k_128": 0.407, "prosst_k_512": 0.401, "prosst_k_1024": 0.327, "prosst_k_2048": 0.433, "prosst_k_4096": 0.334, "escott": 0.5, "venusrem": 0.423, "rsalor": 0.388, "s2f": 0.427, "s2f_msa": 0.429, "s3f": 0.444, "s3f_msa": 0.441, "siterm": 0.433, "esm3_open_1_4b": 0.377, "esm_c_300m": 0.517, "esm_c_600m": 0.471, "xtrimopglm_1b_mlm": 0.317, "xtrimopglm_3b_mlm": 0.305, "xtrimopglm_10b_mlm": 0.369, "xtrimopglm_1b_clm": 0.26, "xtrimopglm_3b_clm": 0.248, "xtrimopglm_7b_clm": 0.116, "xtrimopglm_100b_int4": 0.291, "progen3_112m": 0.345, "progen3_219m": 0.38, "progen3_339m": 0.375, "progen3_762m": 0.35, "progen3_1b": 0.333, "progen3_3b": 0.35, "aido_protein_rag_16b": 0.412, "protriever": 0.442}}, {"id": "AMFR_HUMAN_Tsuboyama_2023_4G3O", "label": "AMFR_HUMAN_Tsuboyama_2023_4G3O", "protein": "AMFR_HUMAN", "mean_mutations": 1.7240915208613727, "n_variants": 2972, "n_graph_configs": 2972, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6068731956473462, "epistasis.sign": 0.3126187479951637, "epistasis.reciprocal_sign": 0.08050805635749007, "diminishing_returns_index": -0.3147096203555029, "increasing_costs_index": 0.29838100777905996, "global_idiosyncratic_index": 0.3633301310269376, "r_s_ratio": 0.6093024086180028, "gamma": 0.7094942752304167, "gamma_star": 0.5263651392898562, "local_optima_ratio": 0.02086137281292059, "autocorrelation": 0.5696023085729107, "fdc": -0.41404332643385094, "evolvability_enhancing_fraction": 0.009509289929146467, "global_optima_accessibility": 0.00639300134589502}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.358, "evmutation": 0.418, "deepsequence_single": 0.419, "deepsequence_ensemble": 0.407, "eve_single": 0.386, "eve_ensemble": 0.398, "unirep": -0.301, "unirep_evotuned": 0.13, "msa_transformer_single": 0.22, "msa_transformer_ensemble": 0.24, "esm_1b": 0.374, "esm_1v_single": 0.07, "esm_1v_ensemble": 0.423, "esm2_8m": -0.209, "esm2_35m": 0.286, "esm2_150m": 0.321, "esm2_650m": 0.261, "esm2_3b": 0.297, "esm2_15b": 0.26, "wavenet": 0.062, "rita_s": -0.052, "rita_m": -0.218, "rita_l": -0.051, "rita_xl": -0.038, "progen2_s": -0.317, "progen2_m": -0.089, "progen2_base": -0.05, "progen2_l": -0.121, "progen2_xl": 0.056, "gemme": 0.396, "vespa": 0.254, "vespal": 0.228, "vespag": 0.289, "protgpt2": 0.09, "tranception_s_no_retrieval": -0.287, "tranception_m_no_retrieval": -0.139, "tranception_l_no_retrieval": -0.275, "tranception_s": 0.392, "tranception_m": 0.402, "tranception_l": 0.356, "trancepteve_s": 0.424, "trancepteve_m": 0.422, "trancepteve_l": 0.374, "carp_38m": -0.185, "carp_600k": -0.194, "carp_640m": 0.165, "carp_76m": 0.114, "mif": 0.251, "mif_st": 0.112, "esm_if1": 0.433, "proteinmpnn": 0.353, "protssn_k_10_h_512": 0.446, "protssn_k_10_h_768": 0.326, "protssn_k_10_h_1280": 0.204, "protssn_k_20_h_512": 0.349, "protssn_k_20_h_768": 0.342, "protssn_k_20_h_1280": 0.349, "protssn_k_30_h_512": 0.343, "protssn_k_30_h_768": 0.339, "protssn_k_30_h_1280": 0.361, "protssn_ensemble": 0.343, "saprot_650m": 0.347, "saprot_35m": 0.397, "poet_200m": 0.272, "mulan": 0.09, "prosst_k_20": 0.363, "prosst_k_128": 0.418, "prosst_k_512": 0.479, "prosst_k_1024": 0.535, "prosst_k_2048": 0.481, "prosst_k_4096": 0.587, "escott": 0.41, "venusrem": 0.496, "rsalor": 0.341, "s2f": 0.328, "s2f_msa": 0.367, "s3f": 0.322, "s3f_msa": 0.362, "siterm": 0.251, "esm3_open_1_4b": 0.29, "esm_c_300m": 0.364, "esm_c_600m": 0.262, "xtrimopglm_1b_mlm": 0.205, "xtrimopglm_3b_mlm": 0.247, "xtrimopglm_10b_mlm": 0.253, "xtrimopglm_1b_clm": -0.064, "xtrimopglm_3b_clm": -0.224, "xtrimopglm_7b_clm": -0.144, "xtrimopglm_100b_int4": 0.262, "progen3_112m": -0.257, "progen3_219m": -0.164, "progen3_339m": -0.253, "progen3_762m": -0.209, "progen3_1b": -0.153, "progen3_3b": -0.133, "aido_protein_rag_16b": 0.445, "protriever": 0.355}}, {"id": "CAPSD_AAV2S_Sinai_2021", "label": "CAPSD_AAV2S_Sinai_2021", "protein": "CAPSD_AAV2S", "mean_mutations": 4.728217728217728, "n_variants": 42328, "n_graph_configs": 28265, "n_isolated_removed": 14063, "features": {"epistasis.magnitude": 0.4436274888398029, "epistasis.sign": 0.36716662290118784, "epistasis.reciprocal_sign": 0.18920588825900925, "diminishing_returns_index": -0.3913836689055911, "increasing_costs_index": 0.16126335218461582, "global_idiosyncratic_index": 0.44578219241219547, "r_s_ratio": 1.96618600466329, "gamma": 0.477815199510688, "gamma_star": 0.25441458696147445, "local_optima_ratio": 0.15934901822041395, "autocorrelation": 0.7027358933179201, "fdc": -0.21925952789974396, "evolvability_enhancing_fraction": 0.0008292842459353392, "global_optima_accessibility": 0.5008314169467539}, "publication": {"title": "Generative AAV capsid diversification by latent interpolation", "journal": "bioRxiv", "year": 2021, "doi": "10.1101/2021.04.16.440236", "url": "https://doi.org/10.1101/2021.04.16.440236"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.407, "evmutation": 0.345, "deepsequence_single": 0.317, "deepsequence_ensemble": 0.372, "eve_single": 0.346, "eve_ensemble": 0.33, "unirep": 0.374, "unirep_evotuned": 0.424, "msa_transformer_single": 0.329, "msa_transformer_ensemble": 0.363, "esm_1b": 0.183, "esm_1v_single": 0.196, "esm_1v_ensemble": 0.199, "esm2_8m": 0.254, "esm2_35m": 0.286, "esm2_150m": 0.203, "esm2_650m": 0.277, "esm2_3b": 0.183, "esm2_15b": 0.124, "wavenet": 0.258, "rita_s": 0.191, "rita_m": 0.25, "rita_l": 0.269, "rita_xl": 0.279, "progen2_s": 0.203, "progen2_m": 0.198, "progen2_base": 0.266, "progen2_l": 0.204, "progen2_xl": 0.399, "gemme": 0.445, "vespa": 0.183, "vespal": 0.177, "vespag": 0.207, "protgpt2": 0.132, "tranception_s_no_retrieval": 0.202, "tranception_m_no_retrieval": 0.262, "tranception_l_no_retrieval": 0.492, "tranception_s": 0.38, "tranception_m": 0.386, "tranception_l": 0.473, "trancepteve_s": 0.338, "trancepteve_m": 0.343, "trancepteve_l": 0.426, "carp_38m": 0.092, "carp_600k": 0.16, "carp_640m": 0.26, "carp_76m": 0.112, "mif": 0.402, "mif_st": 0.393, "esm_if1": 0.347, "proteinmpnn": 0.319, "protssn_k_10_h_512": 0.235, "protssn_k_10_h_768": 0.225, "protssn_k_10_h_1280": 0.216, "protssn_k_20_h_512": 0.235, "protssn_k_20_h_768": 0.214, "protssn_k_20_h_1280": 0.215, "protssn_k_30_h_512": 0.22, "protssn_k_30_h_768": 0.208, "protssn_k_30_h_1280": 0.228, "protssn_ensemble": 0.223, "saprot_650m": 0.301, "saprot_35m": 0.222, "poet_200m": 0.572, "mulan": 0.036, "prosst_k_20": 0.378, "prosst_k_128": 0.571, "prosst_k_512": 0.42, "prosst_k_1024": 0.561, "prosst_k_2048": 0.34, "prosst_k_4096": 0.592, "escott": 0.514, "venusrem": 0.392, "rsalor": 0.555, "s2f": 0.404, "s2f_msa": 0.437, "s3f": 0.421, "s3f_msa": 0.427, "siterm": 0.494, "esm3_open_1_4b": 0.372, "esm_c_300m": 0.21, "esm_c_600m": 0.174, "xtrimopglm_1b_mlm": 0.212, "xtrimopglm_3b_mlm": 0.244, "xtrimopglm_10b_mlm": 0.207, "xtrimopglm_1b_clm": 0.101, "xtrimopglm_3b_clm": 0.069, "xtrimopglm_7b_clm": 0.241, "xtrimopglm_100b_int4": 0.305, "progen3_112m": 0.197, "progen3_219m": 0.276, "progen3_339m": 0.224, "progen3_762m": 0.197, "progen3_1b": 0.22, "progen3_3b": 0.232, "aido_protein_rag_16b": 0.487, "protriever": 0.577}}, {"id": "CBX4_HUMAN_Tsuboyama_2023_2K28", "label": "CBX4_HUMAN_Tsuboyama_2023_2K28", "protein": "CBX4_HUMAN", "mean_mutations": 1.598159509202454, "n_variants": 2282, "n_graph_configs": 2282, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5789214847759152, "epistasis.sign": 0.3368863325350667, "epistasis.reciprocal_sign": 0.08419218268901814, "diminishing_returns_index": -0.3127870861694043, "increasing_costs_index": 0.31424569963956767, "global_idiosyncratic_index": 0.30744502795503703, "r_s_ratio": 0.5398039293405631, "gamma": 0.7700108647262435, "gamma_star": 0.494729302086897, "local_optima_ratio": 0.021034180543382998, "autocorrelation": 0.6156882954204019, "fdc": -0.4559749656374115, "evolvability_enhancing_fraction": 0.004775655264326966, "global_optima_accessibility": 0.15731814198071867}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.498, "evmutation": 0.548, "deepsequence_single": 0.674, "deepsequence_ensemble": 0.706, "eve_single": 0.633, "eve_ensemble": 0.685, "unirep": -0.266, "unirep_evotuned": 0.611, "msa_transformer_single": 0.623, "msa_transformer_ensemble": 0.613, "esm_1b": 0.615, "esm_1v_single": 0.65, "esm_1v_ensemble": 0.648, "esm2_8m": -0.271, "esm2_35m": 0.735, "esm2_150m": 0.699, "esm2_650m": 0.682, "esm2_3b": 0.609, "esm2_15b": 0.606, "wavenet": 0.648, "rita_s": 0.486, "rita_m": 0.589, "rita_l": 0.581, "rita_xl": 0.587, "progen2_s": 0.553, "progen2_m": 0.577, "progen2_base": 0.589, "progen2_l": 0.531, "progen2_xl": 0.569, "gemme": 0.633, "vespa": 0.667, "vespal": 0.653, "vespag": 0.654, "protgpt2": 0.423, "tranception_s_no_retrieval": 0.425, "tranception_m_no_retrieval": 0.505, "tranception_l_no_retrieval": 0.54, "tranception_s": 0.601, "tranception_m": 0.6, "tranception_l": 0.623, "trancepteve_s": 0.676, "trancepteve_m": 0.698, "trancepteve_l": 0.701, "carp_38m": 0.621, "carp_600k": -0.32, "carp_640m": 0.369, "carp_76m": 0.561, "mif": 0.167, "mif_st": 0.155, "esm_if1": 0.614, "proteinmpnn": 0.588, "protssn_k_10_h_512": 0.695, "protssn_k_10_h_768": 0.626, "protssn_k_10_h_1280": 0.69, "protssn_k_20_h_512": 0.676, "protssn_k_20_h_768": 0.709, "protssn_k_20_h_1280": 0.668, "protssn_k_30_h_512": 0.677, "protssn_k_30_h_768": 0.718, "protssn_k_30_h_1280": 0.664, "protssn_ensemble": 0.695, "saprot_650m": 0.642, "saprot_35m": 0.751, "poet_200m": 0.695, "mulan": -0.065, "prosst_k_20": 0.284, "prosst_k_128": 0.648, "prosst_k_512": 0.677, "prosst_k_1024": 0.675, "prosst_k_2048": 0.608, "prosst_k_4096": 0.607, "escott": 0.616, "venusrem": 0.623, "rsalor": 0.656, "s2f": 0.689, "s2f_msa": 0.699, "s3f": 0.689, "s3f_msa": 0.695, "siterm": 0.639, "esm3_open_1_4b": 0.663, "esm_c_300m": 0.627, "esm_c_600m": 0.611, "xtrimopglm_1b_mlm": 0.666, "xtrimopglm_3b_mlm": 0.657, "xtrimopglm_10b_mlm": 0.619, "xtrimopglm_1b_clm": 0.459, "xtrimopglm_3b_clm": 0.433, "xtrimopglm_7b_clm": 0.369, "xtrimopglm_100b_int4": 0.574, "progen3_112m": -0.289, "progen3_219m": 0.375, "progen3_339m": 0.505, "progen3_762m": 0.574, "progen3_1b": 0.599, "progen3_3b": 0.603, "aido_protein_rag_16b": 0.697, "protriever": 0.679}}, {"id": "CSN4_MOUSE_Tsuboyama_2023_1UFM", "label": "CSN4_MOUSE_Tsuboyama_2023_1UFM", "protein": "CSN4_MOUSE", "mean_mutations": 1.5893778452200304, "n_variants": 3295, "n_graph_configs": 3295, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.740244339808184, "epistasis.sign": 0.23318717678771822, "epistasis.reciprocal_sign": 0.026568483404097783, "diminishing_returns_index": -0.35955404347556436, "increasing_costs_index": 0.28948170043096266, "global_idiosyncratic_index": 0.2226341578672673, "r_s_ratio": 0.36295318941675025, "gamma": 0.8574841004519875, "gamma_star": 0.7136758564040863, "local_optima_ratio": 0.022154779969650987, "autocorrelation": 0.613026487323361, "fdc": -0.3979157657364157, "evolvability_enhancing_fraction": 0.007070314894660482, "global_optima_accessibility": 0.005766312594840668}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.389, "evmutation": 0.504, "deepsequence_single": 0.432, "deepsequence_ensemble": 0.445, "eve_single": 0.47, "eve_ensemble": 0.477, "unirep": 0.26, "unirep_evotuned": 0.4, "msa_transformer_single": 0.461, "msa_transformer_ensemble": 0.461, "esm_1b": 0.481, "esm_1v_single": 0.546, "esm_1v_ensemble": 0.605, "esm2_8m": 0.449, "esm2_35m": 0.686, "esm2_150m": 0.518, "esm2_650m": 0.484, "esm2_3b": 0.423, "esm2_15b": 0.415, "wavenet": 0.406, "rita_s": 0.449, "rita_m": 0.455, "rita_l": 0.502, "rita_xl": 0.514, "progen2_s": 0.443, "progen2_m": 0.587, "progen2_base": 0.53, "progen2_l": 0.504, "progen2_xl": 0.501, "gemme": 0.532, "vespa": 0.401, "vespal": 0.389, "vespag": 0.54, "protgpt2": 0.244, "tranception_s_no_retrieval": 0.341, "tranception_m_no_retrieval": 0.474, "tranception_l_no_retrieval": 0.546, "tranception_s": 0.497, "tranception_m": 0.536, "tranception_l": 0.576, "trancepteve_s": 0.468, "trancepteve_m": 0.513, "trancepteve_l": 0.53, "carp_38m": 0.416, "carp_600k": 0.326, "carp_640m": 0.494, "carp_76m": 0.538, "mif": 0.562, "mif_st": 0.541, "esm_if1": 0.653, "proteinmpnn": 0.62, "protssn_k_10_h_512": 0.531, "protssn_k_10_h_768": 0.52, "protssn_k_10_h_1280": 0.532, "protssn_k_20_h_512": 0.557, "protssn_k_20_h_768": 0.58, "protssn_k_20_h_1280": 0.556, "protssn_k_30_h_512": 0.539, "protssn_k_30_h_768": 0.58, "protssn_k_30_h_1280": 0.548, "protssn_ensemble": 0.556, "saprot_650m": 0.575, "saprot_35m": 0.72, "poet_200m": 0.513, "mulan": 0.574, "prosst_k_20": 0.731, "prosst_k_128": 0.701, "prosst_k_512": 0.704, "prosst_k_1024": 0.684, "prosst_k_2048": 0.752, "prosst_k_4096": 0.641, "escott": 0.585, "venusrem": 0.762, "rsalor": 0.554, "s2f": 0.55, "s2f_msa": 0.585, "s3f": 0.534, "s3f_msa": 0.591, "siterm": 0.463, "esm3_open_1_4b": 0.711, "esm_c_300m": 0.514, "esm_c_600m": 0.537, "xtrimopglm_1b_mlm": 0.549, "xtrimopglm_3b_mlm": 0.405, "xtrimopglm_10b_mlm": 0.463, "xtrimopglm_1b_clm": 0.544, "xtrimopglm_3b_clm": 0.4, "xtrimopglm_7b_clm": -0.095, "xtrimopglm_100b_int4": 0.402, "progen3_112m": 0.448, "progen3_219m": 0.486, "progen3_339m": 0.584, "progen3_762m": 0.609, "progen3_1b": 0.585, "progen3_3b": 0.568, "aido_protein_rag_16b": 0.675, "protriever": 0.552}}, {"id": "D7PM05_CLYGR_Somermeyer_2022", "label": "D7PM05_CLYGR_Somermeyer_2022", "protein": "D7PM05_CLYGR", "mean_mutations": 3.0383846624515605, "n_variants": 24515, "n_graph_configs": 12138, "n_isolated_removed": 12377, "features": {"epistasis.magnitude": 0.6277173913043478, "epistasis.sign": 0.31431159420289856, "epistasis.reciprocal_sign": 0.057971014492753624, "diminishing_returns_index": -0.6215684421468989, "increasing_costs_index": 0.17553448812011643, "global_idiosyncratic_index": 0.4554884995349078, "r_s_ratio": null, "gamma": 0.8610349638762544, "gamma_star": 0.5698993791479341, "local_optima_ratio": 0.07455923545888944, "autocorrelation": 0.44681273638520425, "fdc": -0.14575739212475636, "evolvability_enhancing_fraction": 0.0, "global_optima_accessibility": 0.2974130828802109}, "publication": {"title": "Heterogeneity of the GFP fitness landscape and data-driven protein design", "journal": "eLife", "year": 2022, "doi": "10.7554/eLife.75842", "url": "https://doi.org/10.7554/eLife.75842"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.53, "evmutation": 0.625, "deepsequence_single": 0.613, "deepsequence_ensemble": 0.582, "eve_single": 0.629, "eve_ensemble": 0.631, "unirep": 0.073, "unirep_evotuned": 0.475, "msa_transformer_single": 0.649, "msa_transformer_ensemble": 0.657, "esm_1b": 0.439, "esm_1v_single": 0.066, "esm_1v_ensemble": 0.064, "esm2_8m": 0.068, "esm2_35m": 0.053, "esm2_150m": 0.036, "esm2_650m": 0.05, "esm2_3b": 0.072, "esm2_15b": 0.152, "wavenet": 0.438, "rita_s": 0.039, "rita_m": 0.042, "rita_l": 0.114, "rita_xl": 0.082, "progen2_s": 0.009, "progen2_m": 0.047, "progen2_base": 0.068, "progen2_l": 0.194, "progen2_xl": 0.361, "gemme": 0.623, "vespa": 0.563, "vespal": 0.574, "vespag": 0.597, "protgpt2": 0.016, "tranception_s_no_retrieval": 0.096, "tranception_m_no_retrieval": 0.125, "tranception_l_no_retrieval": 0.156, "tranception_s": 0.536, "tranception_m": 0.536, "tranception_l": 0.532, "trancepteve_s": 0.63, "trancepteve_m": 0.629, "trancepteve_l": 0.618, "carp_38m": 0.004, "carp_600k": 0.024, "carp_640m": 0.009, "carp_76m": 0.028, "mif": 0.292, "mif_st": 0.316, "esm_if1": 0.505, "proteinmpnn": 0.396, "protssn_k_10_h_512": 0.495, "protssn_k_10_h_768": 0.491, "protssn_k_10_h_1280": 0.496, "protssn_k_20_h_512": 0.505, "protssn_k_20_h_768": 0.494, "protssn_k_20_h_1280": 0.502, "protssn_k_30_h_512": 0.494, "protssn_k_30_h_768": 0.49, "protssn_k_30_h_1280": 0.487, "protssn_ensemble": 0.498, "saprot_650m": 0.345, "saprot_35m": 0.208, "poet_200m": 0.687, "mulan": 0.187, "prosst_k_20": 0.59, "prosst_k_128": 0.573, "prosst_k_512": 0.601, "prosst_k_1024": 0.602, "prosst_k_2048": 0.524, "prosst_k_4096": 0.631, "escott": 0.672, "venusrem": 0.579, "rsalor": 0.686, "s2f": 0.331, "s2f_msa": 0.554, "s3f": 0.393, "s3f_msa": 0.584, "siterm": 0.655, "esm3_open_1_4b": 0.477, "esm_c_300m": 0.07, "esm_c_600m": 0.054, "xtrimopglm_1b_mlm": 0.042, "xtrimopglm_3b_mlm": 0.055, "xtrimopglm_10b_mlm": 0.061, "xtrimopglm_1b_clm": 0.076, "xtrimopglm_3b_clm": -0.001, "xtrimopglm_7b_clm": -0.027, "xtrimopglm_100b_int4": 0.483, "progen3_112m": 0.034, "progen3_219m": 0.079, "progen3_339m": 0.071, "progen3_762m": 0.079, "progen3_1b": 0.039, "progen3_3b": 0.142, "aido_protein_rag_16b": 0.664, "protriever": 0.674}}, {"id": "DLG4_HUMAN_Faure_2021", "label": "DLG4_HUMAN_Faure_2021", "protein": "DLG4_HUMAN", "mean_mutations": 1.81651376146789, "n_variants": 6976, "n_graph_configs": 6976, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.4906200168839696, "epistasis.sign": 0.38251571147171937, "epistasis.reciprocal_sign": 0.12686427164431105, "diminishing_returns_index": -0.39121193435061963, "increasing_costs_index": 0.24857460982906776, "global_idiosyncratic_index": 0.553501399487907, "r_s_ratio": 0.9238048763830777, "gamma": 0.6641184609120363, "gamma_star": 0.36356179472456907, "local_optima_ratio": 0.03999426605504587, "autocorrelation": 0.5770342379225188, "fdc": -0.2582547730627573, "evolvability_enhancing_fraction": 3.1432702583768155e-05, "global_optima_accessibility": 0.7977350917431193}, "publication": {"title": "Mapping the energetic and allosteric landscapes of protein binding domains", "journal": "Nature", "year": 2022, "doi": "10.1038/s41586-022-04586-4", "url": "https://doi.org/10.1038/s41586-022-04586-4"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.679, "evmutation": 0.584, "deepsequence_single": 0.607, "deepsequence_ensemble": 0.577, "eve_single": 0.609, "eve_ensemble": 0.616, "unirep": 0.723, "unirep_evotuned": 0.629, "msa_transformer_single": 0.524, "msa_transformer_ensemble": 0.535, "esm_1b": 0.52, "esm_1v_single": 0.55, "esm_1v_ensemble": 0.608, "esm2_8m": 0.743, "esm2_35m": 0.764, "esm2_150m": 0.728, "esm2_650m": 0.584, "esm2_3b": 0.487, "esm2_15b": 0.443, "wavenet": 0.65, "rita_s": 0.574, "rita_m": 0.583, "rita_l": 0.569, "rita_xl": 0.538, "progen2_s": 0.605, "progen2_m": 0.607, "progen2_base": 0.564, "progen2_l": 0.562, "progen2_xl": 0.495, "gemme": 0.614, "vespa": 0.634, "vespal": 0.621, "vespag": 0.683, "protgpt2": 0.511, "tranception_s_no_retrieval": 0.576, "tranception_m_no_retrieval": 0.638, "tranception_l_no_retrieval": 0.58, "tranception_s": 0.653, "tranception_m": 0.7, "tranception_l": 0.667, "trancepteve_s": 0.643, "trancepteve_m": 0.661, "trancepteve_l": 0.636, "carp_38m": 0.581, "carp_600k": 0.233, "carp_640m": 0.372, "carp_76m": 0.528, "mif": 0.637, "mif_st": 0.453, "esm_if1": 0.672, "proteinmpnn": 0.315, "protssn_k_10_h_512": 0.483, "protssn_k_10_h_768": 0.37, "protssn_k_10_h_1280": 0.539, "protssn_k_20_h_512": 0.499, "protssn_k_20_h_768": 0.52, "protssn_k_20_h_1280": 0.492, "protssn_k_30_h_512": 0.508, "protssn_k_30_h_768": 0.486, "protssn_k_30_h_1280": 0.504, "protssn_ensemble": 0.504, "saprot_650m": 0.521, "saprot_35m": 0.745, "poet_200m": 0.735, "mulan": 0.689, "prosst_k_20": 0.674, "prosst_k_128": 0.663, "prosst_k_512": 0.673, "prosst_k_1024": 0.685, "prosst_k_2048": 0.739, "prosst_k_4096": 0.705, "escott": 0.635, "venusrem": 0.722, "rsalor": 0.649, "s2f": 0.554, "s2f_msa": 0.607, "s3f": 0.585, "s3f_msa": 0.625, "siterm": 0.614, "esm3_open_1_4b": 0.681, "esm_c_300m": 0.557, "esm_c_600m": 0.562, "xtrimopglm_1b_mlm": 0.54, "xtrimopglm_3b_mlm": 0.517, "xtrimopglm_10b_mlm": 0.51, "xtrimopglm_1b_clm": 0.524, "xtrimopglm_3b_clm": 0.551, "xtrimopglm_7b_clm": 0.456, "xtrimopglm_100b_int4": 0.324, "progen3_112m": 0.67, "progen3_219m": 0.611, "progen3_339m": 0.605, "progen3_762m": 0.676, "progen3_1b": 0.633, "progen3_3b": 0.57, "aido_protein_rag_16b": 0.742, "protriever": 0.699}}, {"id": "DOCK1_MOUSE_Tsuboyama_2023_2M0Y", "label": "DOCK1_MOUSE_Tsuboyama_2023_2M0Y", "protein": "DOCK1_MOUSE", "mean_mutations": 1.583876500857633, "n_variants": 2915, "n_graph_configs": 2915, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.4765004556698347, "epistasis.sign": 0.38052987892201534, "epistasis.reciprocal_sign": 0.14296966540814998, "diminishing_returns_index": -0.19244405932128925, "increasing_costs_index": 0.4451473487968558, "global_idiosyncratic_index": 0.36909573565508424, "r_s_ratio": 0.4322746986410962, "gamma": 0.5371171783520686, "gamma_star": 0.3335307902616847, "local_optima_ratio": 0.026758147512864493, "autocorrelation": 0.5802886018503937, "fdc": -0.3331741612895224, "evolvability_enhancing_fraction": 0.016348146442551233, "global_optima_accessibility": 0.006518010291595197}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.418, "evmutation": 0.453, "deepsequence_single": 0.421, "deepsequence_ensemble": 0.439, "eve_single": 0.494, "eve_ensemble": 0.501, "unirep": 0.057, "unirep_evotuned": 0.432, "msa_transformer_single": 0.406, "msa_transformer_ensemble": 0.403, "esm_1b": 0.525, "esm_1v_single": 0.472, "esm_1v_ensemble": 0.501, "esm2_8m": 0.224, "esm2_35m": 0.458, "esm2_150m": 0.486, "esm2_650m": 0.531, "esm2_3b": 0.508, "esm2_15b": 0.531, "wavenet": 0.518, "rita_s": 0.34, "rita_m": 0.418, "rita_l": 0.402, "rita_xl": 0.408, "progen2_s": 0.315, "progen2_m": 0.402, "progen2_base": 0.365, "progen2_l": 0.348, "progen2_xl": 0.395, "gemme": 0.448, "vespa": 0.471, "vespal": 0.469, "vespag": 0.492, "protgpt2": 0.216, "tranception_s_no_retrieval": 0.202, "tranception_m_no_retrieval": 0.321, "tranception_l_no_retrieval": 0.252, "tranception_s": 0.473, "tranception_m": 0.49, "tranception_l": 0.494, "trancepteve_s": 0.489, "trancepteve_m": 0.491, "trancepteve_l": 0.507, "carp_38m": 0.359, "carp_600k": -0.177, "carp_640m": 0.513, "carp_76m": 0.439, "mif": 0.38, "mif_st": 0.394, "esm_if1": 0.419, "proteinmpnn": 0.439, "protssn_k_10_h_512": 0.492, "protssn_k_10_h_768": 0.481, "protssn_k_10_h_1280": 0.497, "protssn_k_20_h_512": 0.507, "protssn_k_20_h_768": 0.491, "protssn_k_20_h_1280": 0.512, "protssn_k_30_h_512": 0.477, "protssn_k_30_h_768": 0.493, "protssn_k_30_h_1280": 0.504, "protssn_ensemble": 0.497, "saprot_650m": 0.567, "saprot_35m": 0.531, "poet_200m": 0.525, "mulan": 0.159, "prosst_k_20": 0.407, "prosst_k_128": 0.448, "prosst_k_512": 0.557, "prosst_k_1024": 0.559, "prosst_k_2048": 0.594, "prosst_k_4096": 0.573, "escott": 0.55, "venusrem": 0.602, "rsalor": 0.543, "s2f": 0.511, "s2f_msa": 0.53, "s3f": 0.525, "s3f_msa": 0.537, "siterm": 0.495, "esm3_open_1_4b": 0.517, "esm_c_300m": 0.477, "esm_c_600m": 0.472, "xtrimopglm_1b_mlm": 0.416, "xtrimopglm_3b_mlm": 0.377, "xtrimopglm_10b_mlm": 0.446, "xtrimopglm_1b_clm": 0.209, "xtrimopglm_3b_clm": 0.327, "xtrimopglm_7b_clm": 0.146, "xtrimopglm_100b_int4": 0.372, "progen3_112m": 0.21, "progen3_219m": 0.187, "progen3_339m": 0.163, "progen3_762m": 0.259, "progen3_1b": 0.26, "progen3_3b": 0.199, "aido_protein_rag_16b": 0.518, "protriever": 0.524}}, {"id": "F7YBW8_MESOW_Aakre_2015", "label": "F7YBW8_MESOW_Aakre_2015", "protein": "F7YBW8_MESOW", "mean_mutations": 3.574956483899043, "n_variants": 9192, "n_graph_configs": 9192, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.41679997146408454, "epistasis.sign": 0.3468403333214434, "epistasis.reciprocal_sign": 0.2363596952144721, "diminishing_returns_index": 0.2651213833405749, "increasing_costs_index": 0.8566944570814742, "global_idiosyncratic_index": 0.43862687709889964, "r_s_ratio": 4.767627269697688, "gamma": 0.624915885841452, "gamma_star": 0.16861654531415748, "local_optima_ratio": 0.007832898172323759, "autocorrelation": 0.5130989644513265, "fdc": -0.37356964910142576, "evolvability_enhancing_fraction": 0.0011971839374843652, "global_optima_accessibility": 0.9864012184508268}, "publication": {"title": "Evolving New Protein-Protein Interaction Specificity through Promiscuous Intermediates", "journal": "Cell", "year": 2015, "doi": "10.1016/j.cell.2015.09.055", "url": "https://doi.org/10.1016/j.cell.2015.09.055"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.06, "evmutation": 0.395, "deepsequence_single": 0.395, "deepsequence_ensemble": 0.44, "eve_single": 0.428, "eve_ensemble": 0.43, "unirep": 0.016, "unirep_evotuned": 0.32, "msa_transformer_single": 0.382, "msa_transformer_ensemble": 0.395, "esm_1b": 0.437, "esm_1v_single": 0.382, "esm_1v_ensemble": 0.393, "esm2_8m": -0.091, "esm2_35m": -0.001, "esm2_150m": 0.049, "esm2_650m": 0.383, "esm2_3b": 0.366, "esm2_15b": 0.434, "wavenet": 0.391, "rita_s": -0.075, "rita_m": -0.122, "rita_l": -0.104, "rita_xl": -0.005, "progen2_s": 0.098, "progen2_m": 0.125, "progen2_base": -0.021, "progen2_l": 0.298, "progen2_xl": 0.403, "gemme": 0.375, "vespa": 0.461, "vespal": 0.441, "vespag": 0.431, "protgpt2": 0.037, "tranception_s_no_retrieval": -0.036, "tranception_m_no_retrieval": -0.124, "tranception_l_no_retrieval": 0.433, "tranception_s": 0.062, "tranception_m": 0.011, "tranception_l": 0.425, "trancepteve_s": 0.38, "trancepteve_m": 0.367, "trancepteve_l": 0.444, "carp_38m": -0.064, "carp_600k": -0.081, "carp_640m": 0.199, "carp_76m": -0.048, "mif": -0.008, "mif_st": 0.246, "esm_if1": 0.069, "proteinmpnn": 0.035, "protssn_k_10_h_512": 0.431, "protssn_k_10_h_768": 0.447, "protssn_k_10_h_1280": 0.423, "protssn_k_20_h_512": 0.46, "protssn_k_20_h_768": 0.453, "protssn_k_20_h_1280": 0.457, "protssn_k_30_h_512": 0.43, "protssn_k_30_h_768": 0.443, "protssn_k_30_h_1280": 0.447, "protssn_ensemble": 0.448, "saprot_650m": 0.27, "saprot_35m": -0.126, "poet_200m": 0.423, "mulan": -0.094, "prosst_k_20": 0.434, "prosst_k_128": 0.394, "prosst_k_512": 0.427, "prosst_k_1024": 0.382, "prosst_k_2048": 0.104, "prosst_k_4096": 0.429, "escott": 0.399, "venusrem": 0.095, "rsalor": 0.277, "s2f": 0.171, "s2f_msa": 0.349, "s3f": 0.169, "s3f_msa": 0.348, "siterm": 0.409, "esm3_open_1_4b": 0.292, "esm_c_300m": 0.414, "esm_c_600m": 0.445, "xtrimopglm_1b_mlm": 0.395, "xtrimopglm_3b_mlm": 0.446, "xtrimopglm_10b_mlm": 0.476, "xtrimopglm_1b_clm": 0.005, "xtrimopglm_3b_clm": 0.062, "xtrimopglm_7b_clm": -0.181, "xtrimopglm_100b_int4": 0.436, "progen3_112m": -0.082, "progen3_219m": 0.049, "progen3_339m": 0.038, "progen3_762m": 0.235, "progen3_1b": 0.225, "progen3_3b": 0.384, "aido_protein_rag_16b": 0.396, "protriever": 0.444}}, {"id": "F7YBW8_MESOW_Ding_2023", "label": "F7YBW8_MESOW_Ding_2023", "protein": "F7YBW8_MESOW", "mean_mutations": 5.426407472860388, "n_variants": 7922, "n_graph_configs": 2593, "n_isolated_removed": 5329, "features": {"epistasis.magnitude": 0.453125, "epistasis.sign": 0.34375, "epistasis.reciprocal_sign": 0.203125, "diminishing_returns_index": -0.22709702329387862, "increasing_costs_index": 0.38322014086699113, "global_idiosyncratic_index": 0.5619934106536806, "r_s_ratio": 1.3816970691852688, "gamma": 0.4620266049475439, "gamma_star": 0.25, "local_optima_ratio": 0.3359043578866178, "autocorrelation": 0.6380226247592922, "fdc": -0.6306217182361338, "evolvability_enhancing_fraction": 0.0, "global_optima_accessibility": 0.0015426147319706903}, "publication": {"title": "Protein design using structure-based residue preferences", "journal": "Nature Communications", "year": 2024, "doi": "10.1038/s41467-024-45621-4", "url": "https://doi.org/10.1038/s41467-024-45621-4"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.572, "evmutation": 0.535, "deepsequence_single": 0.535, "deepsequence_ensemble": 0.55, "eve_single": 0.55, "eve_ensemble": 0.545, "unirep": 0.499, "unirep_evotuned": 0.463, "msa_transformer_single": 0.462, "msa_transformer_ensemble": 0.468, "esm_1b": 0.488, "esm_1v_single": 0.403, "esm_1v_ensemble": 0.436, "esm2_8m": 0.486, "esm2_35m": 0.626, "esm2_150m": 0.632, "esm2_650m": 0.446, "esm2_3b": 0.456, "esm2_15b": 0.479, "wavenet": 0.498, "rita_s": 0.459, "rita_m": 0.525, "rita_l": 0.413, "rita_xl": 0.403, "progen2_s": 0.531, "progen2_m": 0.387, "progen2_base": 0.454, "progen2_l": 0.38, "progen2_xl": 0.386, "gemme": 0.533, "vespa": 0.556, "vespal": 0.567, "vespag": 0.604, "protgpt2": 0.269, "tranception_s_no_retrieval": 0.484, "tranception_m_no_retrieval": 0.546, "tranception_l_no_retrieval": 0.577, "tranception_s": 0.566, "tranception_m": 0.601, "tranception_l": 0.598, "trancepteve_s": 0.56, "trancepteve_m": 0.583, "trancepteve_l": 0.564, "carp_38m": 0.364, "carp_600k": 0.239, "carp_640m": 0.309, "carp_76m": 0.337, "mif": 0.367, "mif_st": 0.452, "esm_if1": 0.572, "proteinmpnn": 0.441, "protssn_k_10_h_512": 0.493, "protssn_k_10_h_768": 0.528, "protssn_k_10_h_1280": 0.524, "protssn_k_20_h_512": 0.552, "protssn_k_20_h_768": 0.6, "protssn_k_20_h_1280": 0.529, "protssn_k_30_h_512": 0.51, "protssn_k_30_h_768": 0.546, "protssn_k_30_h_1280": 0.556, "protssn_ensemble": 0.539, "saprot_650m": 0.641, "saprot_35m": 0.497, "poet_200m": 0.498, "mulan": 0.44, "prosst_k_20": 0.691, "prosst_k_128": 0.711, "prosst_k_512": 0.719, "prosst_k_1024": 0.682, "prosst_k_2048": 0.692, "prosst_k_4096": 0.712, "escott": 0.571, "venusrem": 0.685, "rsalor": 0.456, "s2f": 0.436, "s2f_msa": 0.494, "s3f": 0.482, "s3f_msa": 0.52, "siterm": 0.551, "esm3_open_1_4b": 0.643, "esm_c_300m": 0.494, "esm_c_600m": 0.534, "xtrimopglm_1b_mlm": 0.529, "xtrimopglm_3b_mlm": 0.417, "xtrimopglm_10b_mlm": 0.437, "xtrimopglm_1b_clm": 0.356, "xtrimopglm_3b_clm": 0.005, "xtrimopglm_7b_clm": 0.002, "xtrimopglm_100b_int4": 0.522, "progen3_112m": 0.513, "progen3_219m": 0.495, "progen3_339m": 0.417, "progen3_762m": 0.328, "progen3_1b": 0.367, "progen3_3b": 0.356, "aido_protein_rag_16b": 0.462, "protriever": 0.532}}, {"id": "GCN4_YEAST_Staller_2018", "label": "GCN4_YEAST_Staller_2018", "protein": "GCN4_YEAST", "mean_mutations": 17.067854435178166, "n_variants": 2638, "n_graph_configs": 136, "n_isolated_removed": 2502, "features": {"epistasis.magnitude": 0.2727272727272727, "epistasis.sign": 0.4318181818181818, "epistasis.reciprocal_sign": 0.29545454545454547, "diminishing_returns_index": -0.678346001501832, "increasing_costs_index": 0.5327266541421067, "global_idiosyncratic_index": 0.7643712521910883, "r_s_ratio": null, "gamma": -0.0020589767505640308, "gamma_star": -0.022727272727272728, "local_optima_ratio": 0.34558823529411764, "autocorrelation": -0.07306903269642916, "fdc": 0.007944856415473515, "evolvability_enhancing_fraction": 0.0, "global_optima_accessibility": 0.014705882352941176}, "publication": {"title": "A High-Throughput Mutational Scan of an Intrinsically Disordered Acidic Transcriptional Activation Domain", "journal": "Cell Systems", "year": 2018, "doi": "10.1016/j.cels.2018.01.015", "url": "https://doi.org/10.1016/j.cels.2018.01.015"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.253, "evmutation": 0.25, "deepsequence_single": 0.245, "deepsequence_ensemble": 0.248, "eve_single": 0.242, "eve_ensemble": 0.241, "unirep": 0.182, "unirep_evotuned": 0.221, "msa_transformer_single": 0.24, "msa_transformer_ensemble": 0.24, "esm_1b": 0.238, "esm_1v_single": 0.272, "esm_1v_ensemble": 0.284, "esm2_8m": 0.318, "esm2_35m": 0.278, "esm2_150m": 0.266, "esm2_650m": 0.281, "esm2_3b": 0.261, "esm2_15b": 0.248, "wavenet": 0.186, "rita_s": 0.175, "rita_m": 0.189, "rita_l": 0.168, "rita_xl": 0.167, "progen2_s": 0.158, "progen2_m": 0.118, "progen2_base": 0.081, "progen2_l": 0.134, "progen2_xl": 0.164, "gemme": 0.225, "vespa": 0.255, "vespal": 0.247, "vespag": 0.258, "protgpt2": 0.025, "tranception_s_no_retrieval": 0.175, "tranception_m_no_retrieval": 0.169, "tranception_l_no_retrieval": 0.265, "tranception_s": 0.258, "tranception_m": 0.257, "tranception_l": 0.278, "trancepteve_s": 0.258, "trancepteve_m": 0.257, "trancepteve_l": 0.274, "carp_38m": 0.107, "carp_600k": 0.192, "carp_640m": 0.157, "carp_76m": 0.126, "mif": 0.145, "mif_st": 0.183, "esm_if1": 0.201, "proteinmpnn": 0.167, "protssn_k_10_h_512": 0.227, "protssn_k_10_h_768": 0.22, "protssn_k_10_h_1280": 0.217, "protssn_k_20_h_512": 0.214, "protssn_k_20_h_768": 0.226, "protssn_k_20_h_1280": 0.224, "protssn_k_30_h_512": 0.219, "protssn_k_30_h_768": 0.221, "protssn_k_30_h_1280": 0.224, "protssn_ensemble": 0.222, "saprot_650m": 0.221, "saprot_35m": 0.245, "poet_200m": 0.158, "mulan": 0.224, "prosst_k_20": 0.235, "prosst_k_128": 0.247, "prosst_k_512": 0.238, "prosst_k_1024": 0.235, "prosst_k_2048": 0.237, "prosst_k_4096": 0.221, "escott": 0.232, "venusrem": 0.231, "rsalor": 0.218, "s2f": 0.19, "s2f_msa": 0.239, "s3f": 0.204, "s3f_msa": 0.257, "siterm": 0.237, "esm3_open_1_4b": 0.211, "esm_c_300m": 0.24, "esm_c_600m": 0.239, "xtrimopglm_1b_mlm": 0.283, "xtrimopglm_3b_mlm": 0.212, "xtrimopglm_10b_mlm": 0.305, "xtrimopglm_1b_clm": 0.1, "xtrimopglm_3b_clm": -0.012, "xtrimopglm_7b_clm": 0.179, "xtrimopglm_100b_int4": 0.311, "progen3_112m": 0.193, "progen3_219m": 0.194, "progen3_339m": 0.181, "progen3_762m": 0.197, "progen3_1b": 0.13, "progen3_3b": 0.129, "aido_protein_rag_16b": 0.271, "protriever": 0.204}}, {"id": "GFP_AEQVI_Sarkisyan_2016", "label": "GFP_AEQVI_Sarkisyan_2016", "protein": "GFP_AEQVI", "mean_mutations": 3.8780020884093283, "n_variants": 51714, "n_graph_configs": 17603, "n_isolated_removed": 34111, "features": {"epistasis.magnitude": 0.42715128922025475, "epistasis.sign": 0.3749611680646163, "epistasis.reciprocal_sign": 0.19788754271512893, "diminishing_returns_index": -0.7584092475257582, "increasing_costs_index": -0.04233867908901653, "global_idiosyncratic_index": 0.4396365101085963, "r_s_ratio": 0.7552460813739483, "gamma": 0.7301011816591767, "gamma_star": 0.22926374650512582, "local_optima_ratio": 0.18979719366017156, "autocorrelation": 0.5181442194841609, "fdc": -0.10907236251297905, "evolvability_enhancing_fraction": 2.9404845918607387e-05, "global_optima_accessibility": 0.424927569164347}, "publication": {"title": "Local fitness landscape of the green fluorescent protein", "journal": "Nature", "year": 2016, "doi": "10.1038/nature17995", "url": "https://doi.org/10.1038/nature17995"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.649, "evmutation": 0.644, "deepsequence_single": 0.672, "deepsequence_ensemble": 0.673, "eve_single": 0.679, "eve_ensemble": 0.679, "unirep": 0.049, "unirep_evotuned": 0.635, "msa_transformer_single": 0.667, "msa_transformer_ensemble": 0.661, "esm_1b": 0.524, "esm_1v_single": 0.098, "esm_1v_ensemble": 0.103, "esm2_8m": 0.078, "esm2_35m": 0.131, "esm2_150m": 0.103, "esm2_650m": 0.108, "esm2_3b": 0.149, "esm2_15b": 0.29, "wavenet": 0.598, "rita_s": 0.078, "rita_m": 0.105, "rita_l": 0.181, "rita_xl": 0.098, "progen2_s": 0.046, "progen2_m": 0.188, "progen2_base": 0.298, "progen2_l": 0.642, "progen2_xl": 0.647, "gemme": 0.678, "vespa": 0.607, "vespal": 0.615, "vespag": 0.62, "protgpt2": 0.074, "tranception_s_no_retrieval": 0.066, "tranception_m_no_retrieval": 0.182, "tranception_l_no_retrieval": 0.629, "tranception_s": 0.649, "tranception_m": 0.651, "tranception_l": 0.672, "trancepteve_s": 0.683, "trancepteve_m": 0.685, "trancepteve_l": 0.706, "carp_38m": 0.024, "carp_600k": -0.014, "carp_640m": 0.041, "carp_76m": 0.036, "mif": 0.512, "mif_st": 0.508, "esm_if1": 0.713, "proteinmpnn": 0.602, "protssn_k_10_h_512": 0.591, "protssn_k_10_h_768": 0.6, "protssn_k_10_h_1280": 0.61, "protssn_k_20_h_512": 0.617, "protssn_k_20_h_768": 0.609, "protssn_k_20_h_1280": 0.611, "protssn_k_30_h_512": 0.605, "protssn_k_30_h_768": 0.606, "protssn_k_30_h_1280": 0.587, "protssn_ensemble": 0.607, "saprot_650m": 0.623, "saprot_35m": 0.449, "poet_200m": 0.714, "mulan": 0.399, "prosst_k_20": 0.657, "prosst_k_128": 0.692, "prosst_k_512": 0.677, "prosst_k_1024": 0.705, "prosst_k_2048": 0.687, "prosst_k_4096": 0.71, "escott": 0.706, "venusrem": 0.695, "rsalor": 0.737, "s2f": 0.571, "s2f_msa": 0.707, "s3f": 0.653, "s3f_msa": 0.735, "siterm": 0.664, "esm3_open_1_4b": 0.697, "esm_c_300m": 0.118, "esm_c_600m": 0.13, "xtrimopglm_1b_mlm": 0.105, "xtrimopglm_3b_mlm": 0.1, "xtrimopglm_10b_mlm": 0.131, "xtrimopglm_1b_clm": 0.028, "xtrimopglm_3b_clm": -0.165, "xtrimopglm_7b_clm": -0.097, "xtrimopglm_100b_int4": 0.602, "progen3_112m": 0.091, "progen3_219m": 0.139, "progen3_339m": 0.154, "progen3_762m": 0.366, "progen3_1b": 0.631, "progen3_3b": 0.632, "aido_protein_rag_16b": 0.696, "protriever": null}}, {"id": "GRB2_HUMAN_Faure_2021", "label": "GRB2_HUMAN_Faure_2021", "protein": "GRB2_HUMAN", "mean_mutations": 1.9836821008111605, "n_variants": 63366, "n_graph_configs": 63366, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6020313942751616, "epistasis.sign": 0.3125577100646353, "epistasis.reciprocal_sign": 0.08541089566020314, "diminishing_returns_index": -0.14216602512634274, "increasing_costs_index": 0.4695130439039562, "global_idiosyncratic_index": 0.41373723737398055, "r_s_ratio": 1.1018318787871046, "gamma": 0.7880738384915607, "gamma_star": 0.516772242076055, "local_optima_ratio": 0.02106808067417858, "autocorrelation": 0.5772015345054127, "fdc": -0.17902526595569804, "evolvability_enhancing_fraction": 0.0036862425712883705, "global_optima_accessibility": 0.8730391692705868}, "publication": {"title": "Mapping the energetic and allosteric landscapes of protein binding domains", "journal": "Nature", "year": 2022, "doi": "10.1038/s41586-022-04586-4", "url": "https://doi.org/10.1038/s41586-022-04586-4"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.405, "evmutation": 0.521, "deepsequence_single": 0.507, "deepsequence_ensemble": 0.538, "eve_single": 0.54, "eve_ensemble": 0.546, "unirep": 0.486, "unirep_evotuned": 0.45, "msa_transformer_single": 0.539, "msa_transformer_ensemble": 0.491, "esm_1b": 0.533, "esm_1v_single": 0.455, "esm_1v_ensemble": 0.515, "esm2_8m": 0.526, "esm2_35m": 0.598, "esm2_150m": 0.626, "esm2_650m": 0.647, "esm2_3b": 0.535, "esm2_15b": 0.579, "wavenet": 0.534, "rita_s": 0.544, "rita_m": 0.516, "rita_l": 0.514, "rita_xl": 0.453, "progen2_s": 0.532, "progen2_m": 0.509, "progen2_base": 0.446, "progen2_l": 0.524, "progen2_xl": 0.448, "gemme": 0.504, "vespa": 0.428, "vespal": 0.422, "vespag": 0.584, "protgpt2": 0.464, "tranception_s_no_retrieval": 0.536, "tranception_m_no_retrieval": 0.506, "tranception_l_no_retrieval": 0.404, "tranception_s": 0.517, "tranception_m": 0.506, "tranception_l": 0.436, "trancepteve_s": 0.562, "trancepteve_m": 0.561, "trancepteve_l": 0.532, "carp_38m": 0.584, "carp_600k": 0.323, "carp_640m": 0.53, "carp_76m": 0.572, "mif": 0.672, "mif_st": 0.527, "esm_if1": 0.69, "proteinmpnn": 0.516, "protssn_k_10_h_512": 0.621, "protssn_k_10_h_768": 0.649, "protssn_k_10_h_1280": 0.638, "protssn_k_20_h_512": 0.638, "protssn_k_20_h_768": 0.647, "protssn_k_20_h_1280": 0.628, "protssn_k_30_h_512": 0.629, "protssn_k_30_h_768": 0.645, "protssn_k_30_h_1280": 0.625, "protssn_ensemble": 0.65, "saprot_650m": 0.556, "saprot_35m": 0.607, "poet_200m": 0.612, "mulan": 0.556, "prosst_k_20": 0.629, "prosst_k_128": 0.639, "prosst_k_512": 0.68, "prosst_k_1024": 0.691, "prosst_k_2048": 0.718, "prosst_k_4096": 0.715, "escott": 0.548, "venusrem": 0.713, "rsalor": 0.625, "s2f": 0.634, "s2f_msa": 0.615, "s3f": 0.632, "s3f_msa": 0.612, "siterm": 0.466, "esm3_open_1_4b": 0.649, "esm_c_300m": 0.626, "esm_c_600m": 0.542, "xtrimopglm_1b_mlm": 0.62, "xtrimopglm_3b_mlm": 0.493, "xtrimopglm_10b_mlm": 0.453, "xtrimopglm_1b_clm": 0.432, "xtrimopglm_3b_clm": 0.388, "xtrimopglm_7b_clm": 0.392, "xtrimopglm_100b_int4": 0.289, "progen3_112m": 0.449, "progen3_219m": 0.444, "progen3_339m": 0.518, "progen3_762m": 0.421, "progen3_1b": 0.438, "progen3_3b": 0.425, "aido_protein_rag_16b": 0.72, "protriever": null}}, {"id": "HECD1_HUMAN_Tsuboyama_2023_3DKM", "label": "HECD1_HUMAN_Tsuboyama_2023_3DKM", "protein": "HECD1_HUMAN", "mean_mutations": 1.7773003938417473, "n_variants": 5586, "n_graph_configs": 5586, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5700798955411048, "epistasis.sign": 0.3456193197243733, "epistasis.reciprocal_sign": 0.08430078473452185, "diminishing_returns_index": -0.2253061312714314, "increasing_costs_index": 0.3298691300260352, "global_idiosyncratic_index": 0.28893776416190947, "r_s_ratio": 0.7003204769595023, "gamma": 0.7152869639953224, "gamma_star": 0.485779110806583, "local_optima_ratio": 0.014858575008950948, "autocorrelation": 0.6886481814320031, "fdc": -0.45536805057040985, "evolvability_enhancing_fraction": 0.009777383674802818, "global_optima_accessibility": 0.00322234156820623}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.291, "evmutation": 0.37, "deepsequence_single": 0.324, "deepsequence_ensemble": 0.324, "eve_single": 0.332, "eve_ensemble": 0.336, "unirep": 0.055, "unirep_evotuned": 0.234, "msa_transformer_single": 0.359, "msa_transformer_ensemble": 0.37, "esm_1b": 0.303, "esm_1v_single": 0.303, "esm_1v_ensemble": 0.293, "esm2_8m": 0.134, "esm2_35m": 0.206, "esm2_150m": 0.474, "esm2_650m": 0.387, "esm2_3b": 0.34, "esm2_15b": 0.385, "wavenet": 0.337, "rita_s": -0.024, "rita_m": 0.232, "rita_l": 0.302, "rita_xl": 0.264, "progen2_s": 0.16, "progen2_m": 0.294, "progen2_base": 0.308, "progen2_l": 0.247, "progen2_xl": 0.214, "gemme": 0.312, "vespa": 0.332, "vespal": 0.307, "vespag": 0.25, "protgpt2": 0.103, "tranception_s_no_retrieval": 0.135, "tranception_m_no_retrieval": 0.101, "tranception_l_no_retrieval": 0.229, "tranception_s": 0.327, "tranception_m": 0.298, "tranception_l": 0.353, "trancepteve_s": 0.327, "trancepteve_m": 0.341, "trancepteve_l": 0.359, "carp_38m": 0.089, "carp_600k": 0.038, "carp_640m": 0.433, "carp_76m": 0.138, "mif": 0.252, "mif_st": 0.26, "esm_if1": 0.38, "proteinmpnn": 0.321, "protssn_k_10_h_512": 0.368, "protssn_k_10_h_768": 0.408, "protssn_k_10_h_1280": 0.345, "protssn_k_20_h_512": 0.392, "protssn_k_20_h_768": 0.411, "protssn_k_20_h_1280": 0.436, "protssn_k_30_h_512": 0.442, "protssn_k_30_h_768": 0.428, "protssn_k_30_h_1280": 0.451, "protssn_ensemble": 0.417, "saprot_650m": 0.414, "saprot_35m": 0.219, "poet_200m": 0.37, "mulan": 0.265, "prosst_k_20": 0.494, "prosst_k_128": 0.338, "prosst_k_512": 0.268, "prosst_k_1024": 0.217, "prosst_k_2048": 0.262, "prosst_k_4096": 0.361, "escott": 0.345, "venusrem": 0.313, "rsalor": 0.339, "s2f": 0.347, "s2f_msa": 0.347, "s3f": 0.401, "s3f_msa": 0.376, "siterm": 0.336, "esm3_open_1_4b": 0.291, "esm_c_300m": 0.406, "esm_c_600m": 0.403, "xtrimopglm_1b_mlm": 0.262, "xtrimopglm_3b_mlm": 0.395, "xtrimopglm_10b_mlm": 0.359, "xtrimopglm_1b_clm": 0.309, "xtrimopglm_3b_clm": 0.33, "xtrimopglm_7b_clm": 0.103, "xtrimopglm_100b_int4": 0.298, "progen3_112m": 0.184, "progen3_219m": 0.204, "progen3_339m": 0.133, "progen3_762m": 0.265, "progen3_1b": 0.205, "progen3_3b": 0.22, "aido_protein_rag_16b": 0.42, "protriever": 0.404}}, {"id": "HIS7_YEAST_Pokusaeva_2019", "label": "HIS7_YEAST_Pokusaeva_2019", "protein": "HIS7_YEAST", "mean_mutations": 6.964781904998015, "n_variants": 496137, "n_graph_configs": 493435, "n_isolated_removed": 2702, "features": {"epistasis.magnitude": 0.4252194944818374, "epistasis.sign": 0.34186346600267153, "epistasis.reciprocal_sign": 0.23291703951549106, "diminishing_returns_index": -0.8053769107106703, "increasing_costs_index": -0.10599311857242383, "global_idiosyncratic_index": 0.5611057693384699, "r_s_ratio": 0.9078533509195411, "gamma": 0.6002924150485807, "gamma_star": 0.23240325128332656, "local_optima_ratio": 0.041513066563985124, "autocorrelation": 0.4440182936329407, "fdc": -0.2145599946525916, "evolvability_enhancing_fraction": 0.0011340654216295864, "global_optima_accessibility": 0.04260743562981953}, "publication": {"title": "An experimental assay of the interactions of amino acids from orthologous sequences shaping a complex fitness landscape", "journal": "PLOS Genetics", "year": 2019, "doi": "10.1371/journal.pgen.1008079", "url": "https://doi.org/10.1371/journal.pgen.1008079"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.472, "evmutation": 0.542, "deepsequence_single": 0.558, "deepsequence_ensemble": 0.557, "eve_single": 0.533, "eve_ensemble": 0.531, "unirep": 0.146, "unirep_evotuned": 0.273, "msa_transformer_single": 0.468, "msa_transformer_ensemble": 0.508, "esm_1b": 0.472, "esm_1v_single": 0.411, "esm_1v_ensemble": 0.477, "esm2_8m": 0.057, "esm2_35m": 0.113, "esm2_150m": 0.44, "esm2_650m": 0.411, "esm2_3b": 0.475, "esm2_15b": 0.48, "wavenet": 0.454, "rita_s": 0.325, "rita_m": 0.402, "rita_l": 0.433, "rita_xl": 0.478, "progen2_s": 0.404, "progen2_m": 0.485, "progen2_base": 0.464, "progen2_l": 0.516, "progen2_xl": 0.539, "gemme": 0.524, "vespa": 0.4, "vespal": 0.374, "vespag": 0.563, "protgpt2": 0.143, "tranception_s_no_retrieval": 0.397, "tranception_m_no_retrieval": 0.453, "tranception_l_no_retrieval": 0.585, "tranception_s": 0.488, "tranception_m": 0.496, "tranception_l": 0.616, "trancepteve_s": 0.573, "trancepteve_m": 0.548, "trancepteve_l": 0.582, "carp_38m": 0.14, "carp_600k": 0.04, "carp_640m": 0.299, "carp_76m": 0.02, "mif": 0.378, "mif_st": 0.424, "esm_if1": 0.529, "proteinmpnn": 0.392, "protssn_k_10_h_512": 0.444, "protssn_k_10_h_768": 0.35, "protssn_k_10_h_1280": 0.404, "protssn_k_20_h_512": 0.427, "protssn_k_20_h_768": 0.44, "protssn_k_20_h_1280": 0.407, "protssn_k_30_h_512": 0.418, "protssn_k_30_h_768": 0.45, "protssn_k_30_h_1280": 0.431, "protssn_ensemble": 0.424, "saprot_650m": 0.521, "saprot_35m": 0.222, "poet_200m": 0.535, "mulan": 0.399, "prosst_k_20": 0.427, "prosst_k_128": 0.403, "prosst_k_512": 0.423, "prosst_k_1024": 0.471, "prosst_k_2048": 0.553, "prosst_k_4096": 0.448, "escott": 0.567, "venusrem": 0.563, "rsalor": 0.452, "s2f": 0.453, "s2f_msa": 0.521, "s3f": 0.539, "s3f_msa": 0.562, "siterm": 0.403, "esm3_open_1_4b": 0.614, "esm_c_300m": 0.474, "esm_c_600m": 0.486, "xtrimopglm_1b_mlm": 0.487, "xtrimopglm_3b_mlm": 0.46, "xtrimopglm_10b_mlm": 0.33, "xtrimopglm_1b_clm": 0.481, "xtrimopglm_3b_clm": 0.078, "xtrimopglm_7b_clm": -0.047, "xtrimopglm_100b_int4": 0.426, "progen3_112m": 0.442, "progen3_219m": 0.43, "progen3_339m": 0.457, "progen3_762m": 0.407, "progen3_1b": 0.511, "progen3_3b": 0.483, "aido_protein_rag_16b": 0.603, "protriever": null}}, {"id": "ISDH_STAAW_Tsuboyama_2023_2LHR", "label": "ISDH_STAAW_Tsuboyama_2023_2LHR", "protein": "ISDH_STAAW", "mean_mutations": 1.5164609053497942, "n_variants": 1944, "n_graph_configs": 1944, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6297430796532197, "epistasis.sign": 0.31691740796419166, "epistasis.reciprocal_sign": 0.05333951238258865, "diminishing_returns_index": -0.5143857919125034, "increasing_costs_index": 0.21191880426660306, "global_idiosyncratic_index": 0.28688146401015513, "r_s_ratio": 0.3112438135630376, "gamma": 0.829479708715824, "gamma_star": 0.576403567270631, "local_optima_ratio": 0.029835390946502057, "autocorrelation": 0.49394041949621387, "fdc": -0.29238828901410857, "evolvability_enhancing_fraction": 0.009213009213009214, "global_optima_accessibility": 0.00977366255144033}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.094, "evmutation": 0.101, "deepsequence_single": 0.206, "deepsequence_ensemble": 0.197, "eve_single": 0.231, "eve_ensemble": 0.231, "unirep": 0.326, "unirep_evotuned": 0.128, "msa_transformer_single": 0.326, "msa_transformer_ensemble": 0.356, "esm_1b": 0.453, "esm_1v_single": 0.413, "esm_1v_ensemble": 0.434, "esm2_8m": 0.377, "esm2_35m": 0.394, "esm2_150m": 0.449, "esm2_650m": 0.429, "esm2_3b": 0.479, "esm2_15b": 0.45, "wavenet": 0.185, "rita_s": 0.326, "rita_m": 0.199, "rita_l": 0.218, "rita_xl": 0.228, "progen2_s": 0.318, "progen2_m": 0.335, "progen2_base": 0.329, "progen2_l": 0.298, "progen2_xl": 0.36, "gemme": 0.304, "vespa": 0.396, "vespal": 0.334, "vespag": 0.39, "protgpt2": 0.236, "tranception_s_no_retrieval": 0.291, "tranception_m_no_retrieval": 0.249, "tranception_l_no_retrieval": 0.298, "tranception_s": 0.259, "tranception_m": 0.219, "tranception_l": 0.245, "trancepteve_s": 0.263, "trancepteve_m": 0.221, "trancepteve_l": 0.244, "carp_38m": 0.3, "carp_600k": 0.248, "carp_640m": 0.349, "carp_76m": 0.329, "mif": 0.592, "mif_st": 0.526, "esm_if1": 0.519, "proteinmpnn": 0.49, "protssn_k_10_h_512": 0.42, "protssn_k_10_h_768": 0.407, "protssn_k_10_h_1280": 0.429, "protssn_k_20_h_512": 0.428, "protssn_k_20_h_768": 0.427, "protssn_k_20_h_1280": 0.429, "protssn_k_30_h_512": 0.409, "protssn_k_30_h_768": 0.414, "protssn_k_30_h_1280": 0.419, "protssn_ensemble": 0.426, "saprot_650m": 0.608, "saprot_35m": 0.522, "poet_200m": 0.295, "mulan": 0.401, "prosst_k_20": 0.61, "prosst_k_128": 0.698, "prosst_k_512": 0.63, "prosst_k_1024": 0.688, "prosst_k_2048": 0.589, "prosst_k_4096": 0.619, "escott": 0.361, "venusrem": 0.576, "rsalor": 0.406, "s2f": 0.514, "s2f_msa": 0.396, "s3f": 0.569, "s3f_msa": 0.435, "siterm": 0.255, "esm3_open_1_4b": 0.638, "esm_c_300m": 0.471, "esm_c_600m": 0.463, "xtrimopglm_1b_mlm": 0.403, "xtrimopglm_3b_mlm": 0.423, "xtrimopglm_10b_mlm": 0.439, "xtrimopglm_1b_clm": 0.322, "xtrimopglm_3b_clm": 0.25, "xtrimopglm_7b_clm": 0.145, "xtrimopglm_100b_int4": 0.434, "progen3_112m": 0.346, "progen3_219m": 0.348, "progen3_339m": 0.321, "progen3_762m": 0.407, "progen3_1b": 0.251, "progen3_3b": 0.413, "aido_protein_rag_16b": 0.493, "protriever": 0.33}}, {"id": "MYO3_YEAST_Tsuboyama_2023_2BTT", "label": "MYO3_YEAST_Tsuboyama_2023_2BTT", "protein": "MYO3_YEAST", "mean_mutations": 1.7127691841067638, "n_variants": 3297, "n_graph_configs": 3297, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6431557043336651, "epistasis.sign": 0.28325341536234505, "epistasis.reciprocal_sign": 0.07359088030398987, "diminishing_returns_index": -0.19829722355653978, "increasing_costs_index": 0.41144709464090073, "global_idiosyncratic_index": 0.3095025802479656, "r_s_ratio": 0.47421876619463554, "gamma": 0.7681809240754965, "gamma_star": 0.5695648240296752, "local_optima_ratio": 0.01910828025477707, "autocorrelation": 0.5656108962199995, "fdc": -0.4051418387336157, "evolvability_enhancing_fraction": 0.0045377256058836786, "global_optima_accessibility": 0.11373976342129208}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.143, "evmutation": 0.248, "deepsequence_single": 0.334, "deepsequence_ensemble": 0.362, "eve_single": 0.381, "eve_ensemble": 0.412, "unirep": 0.155, "unirep_evotuned": 0.251, "msa_transformer_single": 0.428, "msa_transformer_ensemble": 0.419, "esm_1b": 0.519, "esm_1v_single": 0.482, "esm_1v_ensemble": 0.478, "esm2_8m": 0.171, "esm2_35m": 0.411, "esm2_150m": 0.583, "esm2_650m": 0.588, "esm2_3b": 0.507, "esm2_15b": 0.247, "wavenet": 0.321, "rita_s": 0.351, "rita_m": 0.247, "rita_l": 0.336, "rita_xl": 0.311, "progen2_s": 0.342, "progen2_m": 0.287, "progen2_base": 0.364, "progen2_l": 0.313, "progen2_xl": 0.336, "gemme": 0.283, "vespa": 0.279, "vespal": 0.178, "vespag": 0.37, "protgpt2": 0.126, "tranception_s_no_retrieval": 0.419, "tranception_m_no_retrieval": 0.332, "tranception_l_no_retrieval": 0.282, "tranception_s": 0.382, "tranception_m": 0.329, "tranception_l": 0.292, "trancepteve_s": 0.411, "trancepteve_m": 0.401, "trancepteve_l": 0.382, "carp_38m": 0.112, "carp_600k": 0.045, "carp_640m": 0.431, "carp_76m": 0.294, "mif": 0.392, "mif_st": 0.427, "esm_if1": 0.479, "proteinmpnn": 0.359, "protssn_k_10_h_512": 0.517, "protssn_k_10_h_768": 0.495, "protssn_k_10_h_1280": 0.545, "protssn_k_20_h_512": 0.512, "protssn_k_20_h_768": 0.525, "protssn_k_20_h_1280": 0.529, "protssn_k_30_h_512": 0.513, "protssn_k_30_h_768": 0.519, "protssn_k_30_h_1280": 0.506, "protssn_ensemble": 0.525, "saprot_650m": 0.571, "saprot_35m": 0.5, "poet_200m": 0.504, "mulan": 0.3, "prosst_k_20": 0.408, "prosst_k_128": 0.503, "prosst_k_512": 0.575, "prosst_k_1024": 0.589, "prosst_k_2048": 0.688, "prosst_k_4096": 0.585, "escott": 0.284, "venusrem": 0.641, "rsalor": 0.276, "s2f": 0.522, "s2f_msa": 0.491, "s3f": 0.51, "s3f_msa": 0.486, "siterm": 0.228, "esm3_open_1_4b": 0.534, "esm_c_300m": 0.63, "esm_c_600m": 0.618, "xtrimopglm_1b_mlm": 0.471, "xtrimopglm_3b_mlm": 0.517, "xtrimopglm_10b_mlm": 0.55, "xtrimopglm_1b_clm": 0.362, "xtrimopglm_3b_clm": 0.225, "xtrimopglm_7b_clm": 0.177, "xtrimopglm_100b_int4": 0.5, "progen3_112m": 0.178, "progen3_219m": 0.249, "progen3_339m": 0.199, "progen3_762m": 0.326, "progen3_1b": 0.209, "progen3_3b": 0.269, "aido_protein_rag_16b": 0.542, "protriever": 0.467}}, {"id": "NKX31_HUMAN_Tsuboyama_2023_2L9R", "label": "NKX31_HUMAN_Tsuboyama_2023_2L9R", "protein": "NKX31_HUMAN", "mean_mutations": 1.5370668815471393, "n_variants": 2482, "n_graph_configs": 2482, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5020736386643808, "epistasis.sign": 0.3524940811494816, "epistasis.reciprocal_sign": 0.14543228018613763, "diminishing_returns_index": -0.34611251414159205, "increasing_costs_index": 0.2702742903846259, "global_idiosyncratic_index": 0.3840079823659914, "r_s_ratio": 0.5181161758881526, "gamma": 0.5214608024249648, "gamma_star": 0.35664135847824313, "local_optima_ratio": 0.029008863819500404, "autocorrelation": 0.6290391708330755, "fdc": -0.505867354927934, "evolvability_enhancing_fraction": 0.015679345557750635, "global_optima_accessibility": 0.007655116841257051}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.428, "evmutation": 0.548, "deepsequence_single": 0.586, "deepsequence_ensemble": 0.584, "eve_single": 0.607, "eve_ensemble": 0.619, "unirep": 0.582, "unirep_evotuned": 0.553, "msa_transformer_single": 0.584, "msa_transformer_ensemble": 0.608, "esm_1b": 0.56, "esm_1v_single": 0.568, "esm_1v_ensemble": 0.572, "esm2_8m": 0.627, "esm2_35m": 0.638, "esm2_150m": 0.674, "esm2_650m": 0.651, "esm2_3b": 0.582, "esm2_15b": 0.602, "wavenet": 0.558, "rita_s": 0.564, "rita_m": 0.557, "rita_l": 0.556, "rita_xl": 0.564, "progen2_s": 0.576, "progen2_m": 0.579, "progen2_base": 0.606, "progen2_l": 0.54, "progen2_xl": 0.577, "gemme": 0.64, "vespa": 0.525, "vespal": 0.506, "vespag": 0.619, "protgpt2": 0.551, "tranception_s_no_retrieval": 0.494, "tranception_m_no_retrieval": 0.565, "tranception_l_no_retrieval": 0.577, "tranception_s": 0.578, "tranception_m": 0.624, "tranception_l": 0.628, "trancepteve_s": 0.601, "trancepteve_m": 0.618, "trancepteve_l": 0.621, "carp_38m": 0.454, "carp_600k": 0.395, "carp_640m": 0.342, "carp_76m": 0.408, "mif": 0.589, "mif_st": 0.401, "esm_if1": 0.643, "proteinmpnn": 0.642, "protssn_k_10_h_512": 0.634, "protssn_k_10_h_768": 0.626, "protssn_k_10_h_1280": 0.639, "protssn_k_20_h_512": 0.648, "protssn_k_20_h_768": 0.669, "protssn_k_20_h_1280": 0.641, "protssn_k_30_h_512": 0.635, "protssn_k_30_h_768": 0.629, "protssn_k_30_h_1280": 0.642, "protssn_ensemble": 0.645, "saprot_650m": 0.566, "saprot_35m": 0.596, "poet_200m": 0.618, "mulan": 0.456, "prosst_k_20": 0.704, "prosst_k_128": 0.673, "prosst_k_512": 0.669, "prosst_k_1024": 0.586, "prosst_k_2048": 0.655, "prosst_k_4096": 0.706, "escott": 0.66, "venusrem": 0.654, "rsalor": 0.597, "s2f": 0.608, "s2f_msa": 0.641, "s3f": 0.623, "s3f_msa": 0.646, "siterm": 0.591, "esm3_open_1_4b": 0.633, "esm_c_300m": 0.54, "esm_c_600m": 0.556, "xtrimopglm_1b_mlm": 0.543, "xtrimopglm_3b_mlm": 0.534, "xtrimopglm_10b_mlm": 0.55, "xtrimopglm_1b_clm": 0.499, "xtrimopglm_3b_clm": 0.338, "xtrimopglm_7b_clm": 0.45, "xtrimopglm_100b_int4": 0.521, "progen3_112m": 0.537, "progen3_219m": 0.637, "progen3_339m": 0.641, "progen3_762m": 0.593, "progen3_1b": 0.56, "progen3_3b": 0.542, "aido_protein_rag_16b": 0.704, "protriever": 0.566}}, {"id": "OBSCN_HUMAN_Tsuboyama_2023_1V1C", "label": "OBSCN_HUMAN_Tsuboyama_2023_1V1C", "protein": "OBSCN_HUMAN", "mean_mutations": 1.6205817954332187, "n_variants": 3197, "n_graph_configs": 3197, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6308684155797855, "epistasis.sign": 0.29908196916129415, "epistasis.reciprocal_sign": 0.07004961525892035, "diminishing_returns_index": -0.25859892650046523, "increasing_costs_index": 0.26346898844449274, "global_idiosyncratic_index": 0.2514317003288801, "r_s_ratio": 0.4512054663938362, "gamma": 0.7220425692232216, "gamma_star": 0.5608188003208652, "local_optima_ratio": 0.022521113543947452, "autocorrelation": 0.7223292497123601, "fdc": -0.5614577416217855, "evolvability_enhancing_fraction": 0.010995177918190888, "global_optima_accessibility": 0.005004691898654989}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.54, "evmutation": 0.697, "deepsequence_single": 0.759, "deepsequence_ensemble": 0.777, "eve_single": 0.745, "eve_ensemble": 0.757, "unirep": 0.384, "unirep_evotuned": 0.455, "msa_transformer_single": 0.751, "msa_transformer_ensemble": 0.763, "esm_1b": 0.691, "esm_1v_single": 0.735, "esm_1v_ensemble": 0.759, "esm2_8m": 0.392, "esm2_35m": 0.779, "esm2_150m": 0.819, "esm2_650m": 0.797, "esm2_3b": 0.786, "esm2_15b": 0.755, "wavenet": 0.723, "rita_s": 0.688, "rita_m": 0.689, "rita_l": 0.663, "rita_xl": 0.687, "progen2_s": 0.589, "progen2_m": 0.667, "progen2_base": 0.694, "progen2_l": 0.726, "progen2_xl": 0.604, "gemme": 0.735, "vespa": 0.694, "vespal": 0.679, "vespag": 0.696, "protgpt2": 0.549, "tranception_s_no_retrieval": 0.262, "tranception_m_no_retrieval": 0.461, "tranception_l_no_retrieval": 0.575, "tranception_s": 0.687, "tranception_m": 0.695, "tranception_l": 0.704, "trancepteve_s": 0.755, "trancepteve_m": 0.76, "trancepteve_l": 0.764, "carp_38m": 0.543, "carp_600k": 0.115, "carp_640m": 0.507, "carp_76m": 0.633, "mif": 0.63, "mif_st": 0.54, "esm_if1": 0.782, "proteinmpnn": 0.747, "protssn_k_10_h_512": 0.762, "protssn_k_10_h_768": 0.76, "protssn_k_10_h_1280": 0.761, "protssn_k_20_h_512": 0.755, "protssn_k_20_h_768": 0.772, "protssn_k_20_h_1280": 0.76, "protssn_k_30_h_512": 0.773, "protssn_k_30_h_768": 0.767, "protssn_k_30_h_1280": 0.776, "protssn_ensemble": 0.771, "saprot_650m": 0.825, "saprot_35m": 0.772, "poet_200m": 0.764, "mulan": 0.572, "prosst_k_20": 0.782, "prosst_k_128": 0.806, "prosst_k_512": 0.759, "prosst_k_1024": 0.747, "prosst_k_2048": 0.793, "prosst_k_4096": 0.754, "escott": 0.773, "venusrem": 0.772, "rsalor": 0.775, "s2f": 0.802, "s2f_msa": 0.802, "s3f": 0.824, "s3f_msa": 0.812, "siterm": 0.643, "esm3_open_1_4b": 0.806, "esm_c_300m": 0.82, "esm_c_600m": 0.818, "xtrimopglm_1b_mlm": 0.643, "xtrimopglm_3b_mlm": 0.661, "xtrimopglm_10b_mlm": 0.695, "xtrimopglm_1b_clm": 0.663, "xtrimopglm_3b_clm": 0.395, "xtrimopglm_7b_clm": 0.146, "xtrimopglm_100b_int4": 0.603, "progen3_112m": 0.473, "progen3_219m": 0.394, "progen3_339m": 0.476, "progen3_762m": 0.425, "progen3_1b": 0.55, "progen3_3b": 0.643, "aido_protein_rag_16b": 0.839, "protriever": 0.756}}, {"id": "PABP_YEAST_Melamed_2013", "label": "PABP_YEAST_Melamed_2013", "protein": "PABP_YEAST", "mean_mutations": 1.9685212686962978, "n_variants": 37708, "n_graph_configs": 37708, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5351713743073838, "epistasis.sign": 0.358395229388424, "epistasis.reciprocal_sign": 0.10643339630419224, "diminishing_returns_index": -0.21084069971596586, "increasing_costs_index": 0.44735384425722896, "global_idiosyncratic_index": 0.5006316549452908, "r_s_ratio": 0.8404067437267045, "gamma": 0.7203626959074846, "gamma_star": 0.4287084338384598, "local_optima_ratio": 0.03824122202185213, "autocorrelation": 0.5715928676103502, "fdc": -0.1454260341384879, "evolvability_enhancing_fraction": 0.00019819059328721121, "global_optima_accessibility": 0.2852975495915986}, "publication": {"title": "Deep mutational scanning of an RRM domain of the Saccharomyces cerevisiae poly(A)-binding protein", "journal": "RNA", "year": 2013, "doi": "10.1261/rna.040709.113", "url": "https://doi.org/10.1261/rna.040709.113"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.663, "evmutation": 0.617, "deepsequence_single": 0.541, "deepsequence_ensemble": 0.55, "eve_single": 0.654, "eve_ensemble": 0.648, "unirep": 0.474, "unirep_evotuned": 0.569, "msa_transformer_single": 0.637, "msa_transformer_ensemble": 0.663, "esm_1b": 0.688, "esm_1v_single": 0.662, "esm_1v_ensemble": 0.678, "esm2_8m": 0.476, "esm2_35m": 0.566, "esm2_150m": 0.648, "esm2_650m": 0.716, "esm2_3b": 0.684, "esm2_15b": 0.695, "wavenet": 0.541, "rita_s": 0.638, "rita_m": 0.665, "rita_l": 0.666, "rita_xl": 0.692, "progen2_s": 0.638, "progen2_m": 0.698, "progen2_base": 0.7, "progen2_l": 0.676, "progen2_xl": 0.666, "gemme": 0.674, "vespa": 0.703, "vespal": 0.635, "vespag": 0.671, "protgpt2": 0.261, "tranception_s_no_retrieval": 0.638, "tranception_m_no_retrieval": 0.648, "tranception_l_no_retrieval": 0.64, "tranception_s": 0.689, "tranception_m": 0.692, "tranception_l": 0.688, "trancepteve_s": 0.683, "trancepteve_m": 0.687, "trancepteve_l": 0.684, "carp_38m": 0.605, "carp_600k": -0.027, "carp_640m": 0.67, "carp_76m": 0.654, "mif": 0.361, "mif_st": 0.612, "esm_if1": 0.514, "proteinmpnn": 0.197, "protssn_k_10_h_512": 0.664, "protssn_k_10_h_768": 0.66, "protssn_k_10_h_1280": 0.672, "protssn_k_20_h_512": 0.693, "protssn_k_20_h_768": 0.685, "protssn_k_20_h_1280": 0.689, "protssn_k_30_h_512": 0.691, "protssn_k_30_h_768": 0.69, "protssn_k_30_h_1280": 0.718, "protssn_ensemble": 0.705, "saprot_650m": 0.715, "saprot_35m": 0.622, "poet_200m": 0.689, "mulan": 0.544, "prosst_k_20": 0.638, "prosst_k_128": 0.68, "prosst_k_512": 0.698, "prosst_k_1024": 0.716, "prosst_k_2048": 0.716, "prosst_k_4096": 0.724, "escott": 0.7, "venusrem": 0.712, "rsalor": 0.644, "s2f": 0.694, "s2f_msa": 0.715, "s3f": 0.735, "s3f_msa": 0.733, "siterm": 0.627, "esm3_open_1_4b": 0.647, "esm_c_300m": 0.715, "esm_c_600m": 0.727, "xtrimopglm_1b_mlm": 0.678, "xtrimopglm_3b_mlm": 0.689, "xtrimopglm_10b_mlm": 0.734, "xtrimopglm_1b_clm": 0.658, "xtrimopglm_3b_clm": 0.351, "xtrimopglm_7b_clm": 0.316, "xtrimopglm_100b_int4": 0.675, "progen3_112m": 0.641, "progen3_219m": 0.626, "progen3_339m": 0.65, "progen3_762m": 0.66, "progen3_1b": 0.689, "progen3_3b": 0.686, "aido_protein_rag_16b": 0.723, "protriever": 0.677}}, {"id": "PHOT_CHLRE_Chen_2023", "label": "PHOT_CHLRE_Chen_2023", "protein": "PHOT_CHLRE", "mean_mutations": 7.984277348996293, "n_variants": 167529, "n_graph_configs": 167529, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5738178562708554, "epistasis.sign": 0.3323721936076457, "epistasis.reciprocal_sign": 0.09380995012149891, "diminishing_returns_index": -0.12974444863629808, "increasing_costs_index": 0.3484573483614171, "global_idiosyncratic_index": 0.26205416677013077, "r_s_ratio": 0.483061659955033, "gamma": 0.796376300017953, "gamma_star": 0.48212119884435073, "local_optima_ratio": 0.003945585540413898, "autocorrelation": 0.7341981669368188, "fdc": -0.23151001262035117, "evolvability_enhancing_fraction": 0.00034828679661495435, "global_optima_accessibility": 0.929797229136448}, "publication": {"title": "Deep Mutational Scanning of an Oxygen-Independent Fluorescent Protein CreiLOV for Comprehensive Profiling of Mutational and Epistatic Effects", "journal": "ACS Synthetic Biology", "year": 2023, "doi": "10.1021/acssynbio.2c00662", "url": "https://doi.org/10.1021/acssynbio.2c00662"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.211, "evmutation": 0.44, "deepsequence_single": 0.731, "deepsequence_ensemble": 0.706, "eve_single": 0.352, "eve_ensemble": 0.335, "unirep": 0.654, "unirep_evotuned": 0.552, "msa_transformer_single": 0.711, "msa_transformer_ensemble": 0.721, "esm_1b": 0.647, "esm_1v_single": 0.757, "esm_1v_ensemble": 0.783, "esm2_8m": 0.768, "esm2_35m": 0.823, "esm2_150m": 0.714, "esm2_650m": 0.749, "esm2_3b": 0.714, "esm2_15b": 0.758, "wavenet": 0.652, "rita_s": 0.685, "rita_m": 0.678, "rita_l": 0.596, "rita_xl": 0.699, "progen2_s": 0.533, "progen2_m": 0.616, "progen2_base": 0.596, "progen2_l": 0.656, "progen2_xl": 0.602, "gemme": 0.587, "vespa": 0.572, "vespal": 0.49, "vespag": 0.596, "protgpt2": 0.303, "tranception_s_no_retrieval": 0.565, "tranception_m_no_retrieval": 0.511, "tranception_l_no_retrieval": 0.602, "tranception_s": 0.597, "tranception_m": 0.564, "tranception_l": 0.606, "trancepteve_s": 0.416, "trancepteve_m": 0.445, "trancepteve_l": 0.422, "carp_38m": 0.623, "carp_600k": 0.395, "carp_640m": 0.599, "carp_76m": 0.621, "mif": 0.2, "mif_st": 0.502, "esm_if1": 0.669, "proteinmpnn": 0.41, "protssn_k_10_h_512": 0.603, "protssn_k_10_h_768": 0.543, "protssn_k_10_h_1280": 0.553, "protssn_k_20_h_512": 0.551, "protssn_k_20_h_768": 0.555, "protssn_k_20_h_1280": 0.57, "protssn_k_30_h_512": 0.568, "protssn_k_30_h_768": 0.574, "protssn_k_30_h_1280": 0.583, "protssn_ensemble": 0.569, "saprot_650m": 0.737, "saprot_35m": 0.775, "poet_200m": 0.702, "mulan": 0.173, "prosst_k_20": 0.309, "prosst_k_128": 0.379, "prosst_k_512": 0.368, "prosst_k_1024": 0.422, "prosst_k_2048": 0.438, "prosst_k_4096": 0.355, "escott": 0.517, "venusrem": 0.445, "rsalor": 0.753, "s2f": 0.75, "s2f_msa": 0.62, "s3f": 0.74, "s3f_msa": 0.62, "siterm": 0.411, "esm3_open_1_4b": 0.652, "esm_c_300m": 0.757, "esm_c_600m": 0.696, "xtrimopglm_1b_mlm": 0.715, "xtrimopglm_3b_mlm": 0.664, "xtrimopglm_10b_mlm": 0.669, "xtrimopglm_1b_clm": 0.578, "xtrimopglm_3b_clm": 0.653, "xtrimopglm_7b_clm": 0.298, "xtrimopglm_100b_int4": 0.522, "progen3_112m": 0.531, "progen3_219m": 0.47, "progen3_339m": 0.523, "progen3_762m": 0.552, "progen3_1b": 0.665, "progen3_3b": 0.722, "aido_protein_rag_16b": 0.697, "protriever": 0.726}}, {"id": "POLG_PESV_Tsuboyama_2023_2MXD", "label": "POLG_PESV_Tsuboyama_2023_2MXD", "protein": "POLG_PESV", "mean_mutations": 1.8060428849902534, "n_variants": 5130, "n_graph_configs": 5130, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.670841385044041, "epistasis.sign": 0.285491191802984, "epistasis.reciprocal_sign": 0.04366742315297501, "diminishing_returns_index": -0.21450629953443742, "increasing_costs_index": 0.3644636968540109, "global_idiosyncratic_index": 0.27674350962343586, "r_s_ratio": 0.5413323748926254, "gamma": 0.8004554815079664, "gamma_star": 0.627173961891066, "local_optima_ratio": 0.010721247563352826, "autocorrelation": 0.6985105228801686, "fdc": -0.48763084062557194, "evolvability_enhancing_fraction": 0.010653935185185185, "global_optima_accessibility": 0.0035087719298245615}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.251, "evmutation": 0.455, "deepsequence_single": 0.373, "deepsequence_ensemble": 0.407, "eve_single": 0.471, "eve_ensemble": 0.463, "unirep": 0.07, "unirep_evotuned": 0.42, "msa_transformer_single": 0.39, "msa_transformer_ensemble": 0.514, "esm_1b": 0.375, "esm_1v_single": 0.062, "esm_1v_ensemble": 0.157, "esm2_8m": 0.132, "esm2_35m": 0.14, "esm2_150m": 0.031, "esm2_650m": 0.16, "esm2_3b": 0.114, "esm2_15b": 0.084, "wavenet": 0.592, "rita_s": 0.079, "rita_m": 0.058, "rita_l": 0.016, "rita_xl": 0.157, "progen2_s": 0.106, "progen2_m": -0.059, "progen2_base": 0.008, "progen2_l": -0.084, "progen2_xl": 0.17, "gemme": 0.507, "vespa": 0.575, "vespal": 0.58, "vespag": 0.557, "protgpt2": 0.131, "tranception_s_no_retrieval": 0.003, "tranception_m_no_retrieval": -0.033, "tranception_l_no_retrieval": 0.053, "tranception_s": 0.382, "tranception_m": 0.383, "tranception_l": 0.384, "trancepteve_s": 0.453, "trancepteve_m": 0.463, "trancepteve_l": 0.452, "carp_38m": 0.006, "carp_600k": -0.037, "carp_640m": 0.033, "carp_76m": 0.026, "mif": 0.448, "mif_st": 0.443, "esm_if1": 0.65, "proteinmpnn": 0.576, "protssn_k_10_h_512": 0.658, "protssn_k_10_h_768": 0.664, "protssn_k_10_h_1280": 0.615, "protssn_k_20_h_512": 0.736, "protssn_k_20_h_768": 0.705, "protssn_k_20_h_1280": 0.718, "protssn_k_30_h_512": 0.692, "protssn_k_30_h_768": 0.731, "protssn_k_30_h_1280": 0.718, "protssn_ensemble": 0.718, "saprot_650m": 0.716, "saprot_35m": 0.686, "poet_200m": 0.561, "mulan": 0.433, "prosst_k_20": 0.794, "prosst_k_128": 0.752, "prosst_k_512": 0.815, "prosst_k_1024": 0.798, "prosst_k_2048": 0.808, "prosst_k_4096": 0.805, "escott": 0.572, "venusrem": 0.759, "rsalor": 0.6, "s2f": 0.592, "s2f_msa": 0.661, "s3f": 0.621, "s3f_msa": 0.687, "siterm": 0.529, "esm3_open_1_4b": 0.625, "esm_c_300m": 0.25, "esm_c_600m": 0.247, "xtrimopglm_1b_mlm": 0.23, "xtrimopglm_3b_mlm": 0.414, "xtrimopglm_10b_mlm": 0.239, "xtrimopglm_1b_clm": -0.122, "xtrimopglm_3b_clm": -0.016, "xtrimopglm_7b_clm": -0.17, "xtrimopglm_100b_int4": 0.269, "progen3_112m": 0.195, "progen3_219m": 0.243, "progen3_339m": 0.077, "progen3_762m": 0.011, "progen3_1b": 0.205, "progen3_3b": 0.215, "aido_protein_rag_16b": 0.651, "protriever": 0.58}}, {"id": "Q6WV12_9MAXI_Somermeyer_2022", "label": "Q6WV12_9MAXI_Somermeyer_2022", "protein": "Q6WV12_9MAXI", "mean_mutations": 2.6851055698863093, "n_variants": 31401, "n_graph_configs": 19302, "n_isolated_removed": 12099, "features": {"epistasis.magnitude": 0.5571776155717761, "epistasis.sign": 0.35658286023249525, "epistasis.reciprocal_sign": 0.08623952419572857, "diminishing_returns_index": -0.5111188825590363, "increasing_costs_index": 0.19726750551189048, "global_idiosyncratic_index": 0.2922030772631801, "r_s_ratio": 0.8704371981051255, "gamma": 0.8461172836101017, "gamma_star": 0.47101743007701663, "local_optima_ratio": 0.12066107139156564, "autocorrelation": 0.5960435532260402, "fdc": -0.15246966598314546, "evolvability_enhancing_fraction": 0.0, "global_optima_accessibility": 0.16884260698373227}, "publication": {"title": "Heterogeneity of the GFP fitness landscape and data-driven protein design", "journal": "eLife", "year": 2022, "doi": "10.7554/eLife.75842", "url": "https://doi.org/10.7554/eLife.75842"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.327, "evmutation": 0.393, "deepsequence_single": 0.26, "deepsequence_ensemble": 0.264, "eve_single": 0.333, "eve_ensemble": 0.334, "unirep": -0.031, "unirep_evotuned": 0.115, "msa_transformer_single": 0.413, "msa_transformer_ensemble": 0.407, "esm_1b": 0.242, "esm_1v_single": 0.036, "esm_1v_ensemble": 0.022, "esm2_8m": 0.014, "esm2_35m": 0.046, "esm2_150m": 0.011, "esm2_650m": 0.01, "esm2_3b": 0.01, "esm2_15b": -0.022, "wavenet": 0.294, "rita_s": 0.013, "rita_m": 0.045, "rita_l": 0.003, "rita_xl": 0.102, "progen2_s": 0.004, "progen2_m": -0.002, "progen2_base": 0.03, "progen2_l": -0.005, "progen2_xl": -0.013, "gemme": 0.426, "vespa": 0.344, "vespal": 0.354, "vespag": 0.359, "protgpt2": 0.043, "tranception_s_no_retrieval": 0.031, "tranception_m_no_retrieval": 0.022, "tranception_l_no_retrieval": 0.014, "tranception_s": 0.248, "tranception_m": 0.246, "tranception_l": 0.245, "trancepteve_s": 0.33, "trancepteve_m": 0.327, "trancepteve_l": 0.324, "carp_38m": 0.014, "carp_600k": -0.015, "carp_640m": -0.004, "carp_76m": 0.019, "mif": 0.161, "mif_st": 0.156, "esm_if1": 0.299, "proteinmpnn": 0.169, "protssn_k_10_h_512": 0.228, "protssn_k_10_h_768": 0.248, "protssn_k_10_h_1280": 0.237, "protssn_k_20_h_512": 0.287, "protssn_k_20_h_768": 0.267, "protssn_k_20_h_1280": 0.272, "protssn_k_30_h_512": 0.261, "protssn_k_30_h_768": 0.263, "protssn_k_30_h_1280": 0.264, "protssn_ensemble": 0.261, "saprot_650m": 0.072, "saprot_35m": 0.031, "poet_200m": 0.456, "mulan": 0.052, "prosst_k_20": 0.386, "prosst_k_128": 0.457, "prosst_k_512": 0.444, "prosst_k_1024": 0.41, "prosst_k_2048": 0.265, "prosst_k_4096": 0.466, "escott": 0.467, "venusrem": 0.314, "rsalor": 0.514, "s2f": 0.104, "s2f_msa": 0.273, "s3f": 0.172, "s3f_msa": 0.311, "siterm": 0.39, "esm3_open_1_4b": 0.149, "esm_c_300m": 0.002, "esm_c_600m": -0.007, "xtrimopglm_1b_mlm": 0.042, "xtrimopglm_3b_mlm": -0.012, "xtrimopglm_10b_mlm": -0.003, "xtrimopglm_1b_clm": 0.001, "xtrimopglm_3b_clm": -0.06, "xtrimopglm_7b_clm": 0.048, "xtrimopglm_100b_int4": 0.083, "progen3_112m": -0.014, "progen3_219m": -0.015, "progen3_339m": -0.005, "progen3_762m": 0.016, "progen3_1b": -0.024, "progen3_3b": 0.002, "aido_protein_rag_16b": 0.428, "protriever": 0.477}}, {"id": "Q8WTC7_9CNID_Somermeyer_2022", "label": "Q8WTC7_9CNID_Somermeyer_2022", "protein": "Q8WTC7_9CNID", "mean_mutations": 3.055207400775888, "n_variants": 33510, "n_graph_configs": 19288, "n_isolated_removed": 14222, "features": {"epistasis.magnitude": 0.5382726156365125, "epistasis.sign": 0.364203847728203, "epistasis.reciprocal_sign": 0.09752353663528449, "diminishing_returns_index": -0.5305268573784495, "increasing_costs_index": 0.216571612173795, "global_idiosyncratic_index": 0.29036896526985495, "r_s_ratio": null, "gamma": 0.8543205088993265, "gamma_star": 0.4409050236482168, "local_optima_ratio": 0.1294068851099129, "autocorrelation": 0.5601594267797937, "fdc": -0.047386885158550725, "evolvability_enhancing_fraction": 0.0, "global_optima_accessibility": 0.4737660721692244}, "publication": {"title": "Heterogeneity of the GFP fitness landscape and data-driven protein design", "journal": "eLife", "year": 2022, "doi": "10.7554/eLife.75842", "url": "https://doi.org/10.7554/eLife.75842"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.286, "evmutation": 0.368, "deepsequence_single": 0.217, "deepsequence_ensemble": 0.215, "eve_single": 0.31, "eve_ensemble": 0.305, "unirep": 0.025, "unirep_evotuned": 0.303, "msa_transformer_single": 0.316, "msa_transformer_ensemble": 0.322, "esm_1b": 0.223, "esm_1v_single": -0.009, "esm_1v_ensemble": -0.018, "esm2_8m": -0.027, "esm2_35m": -0.035, "esm2_150m": -0.023, "esm2_650m": -0.025, "esm2_3b": -0.007, "esm2_15b": 0.034, "wavenet": 0.23, "rita_s": 0.013, "rita_m": 0.037, "rita_l": 0.057, "rita_xl": 0.013, "progen2_s": -0.005, "progen2_m": -0.001, "progen2_base": 0.037, "progen2_l": 0.252, "progen2_xl": 0.267, "gemme": 0.373, "vespa": 0.297, "vespal": 0.307, "vespag": 0.299, "protgpt2": -0.018, "tranception_s_no_retrieval": -0.026, "tranception_m_no_retrieval": 0.009, "tranception_l_no_retrieval": 0.303, "tranception_s": 0.231, "tranception_m": 0.237, "tranception_l": 0.318, "trancepteve_s": 0.304, "trancepteve_m": 0.309, "trancepteve_l": 0.345, "carp_38m": -0.013, "carp_600k": -0.008, "carp_640m": -0.024, "carp_76m": -0.027, "mif": 0.097, "mif_st": 0.143, "esm_if1": 0.292, "proteinmpnn": 0.189, "protssn_k_10_h_512": 0.242, "protssn_k_10_h_768": 0.234, "protssn_k_10_h_1280": 0.246, "protssn_k_20_h_512": 0.26, "protssn_k_20_h_768": 0.24, "protssn_k_20_h_1280": 0.254, "protssn_k_30_h_512": 0.249, "protssn_k_30_h_768": 0.243, "protssn_k_30_h_1280": 0.238, "protssn_ensemble": 0.247, "saprot_650m": 0.137, "saprot_35m": -0.001, "poet_200m": 0.353, "mulan": 0.042, "prosst_k_20": 0.355, "prosst_k_128": 0.388, "prosst_k_512": 0.393, "prosst_k_1024": 0.42, "prosst_k_2048": 0.279, "prosst_k_4096": 0.475, "escott": 0.417, "venusrem": 0.315, "rsalor": 0.485, "s2f": 0.162, "s2f_msa": 0.277, "s3f": 0.201, "s3f_msa": 0.302, "siterm": 0.353, "esm3_open_1_4b": 0.211, "esm_c_300m": -0.012, "esm_c_600m": -0.037, "xtrimopglm_1b_mlm": -0.016, "xtrimopglm_3b_mlm": -0.026, "xtrimopglm_10b_mlm": -0.018, "xtrimopglm_1b_clm": 0.004, "xtrimopglm_3b_clm": -0.047, "xtrimopglm_7b_clm": 0.011, "xtrimopglm_100b_int4": 0.299, "progen3_112m": -0.018, "progen3_219m": 0.032, "progen3_339m": -0.001, "progen3_762m": 0.03, "progen3_1b": 0.215, "progen3_3b": 0.247, "aido_protein_rag_16b": 0.32, "protriever": 0.337}}, {"id": "RASK_HUMAN_Weng_2022_abundance", "label": "RASK_HUMAN_Weng_2022_abundance", "protein": "RASK_HUMAN", "mean_mutations": 1.8821313240043056, "n_variants": 26012, "n_graph_configs": 26011, "n_isolated_removed": 1, "features": {"epistasis.magnitude": 0.5326315789473685, "epistasis.sign": 0.3705263157894737, "epistasis.reciprocal_sign": 0.0968421052631579, "diminishing_returns_index": -0.3036878933778092, "increasing_costs_index": 0.3502320801923999, "global_idiosyncratic_index": 0.47904741440126086, "r_s_ratio": null, "gamma": 0.6999035981221501, "gamma_star": 0.4407252440725244, "local_optima_ratio": 0.027257698665949022, "autocorrelation": 0.5291192996231249, "fdc": -0.1233891734986464, "evolvability_enhancing_fraction": 0.00016374490516605312, "global_optima_accessibility": 0.3736496097804775}, "publication": {"title": "The energetic and allosteric landscape for KRAS inhibition", "journal": "Nature", "year": 2024, "doi": "10.1038/s41586-023-06954-0", "url": "https://doi.org/10.1038/s41586-023-06954-0"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.299, "evmutation": 0.244, "deepsequence_single": 0.356, "deepsequence_ensemble": 0.346, "eve_single": 0.332, "eve_ensemble": 0.353, "unirep": 0.211, "unirep_evotuned": 0.244, "msa_transformer_single": 0.244, "msa_transformer_ensemble": 0.271, "esm_1b": 0.276, "esm_1v_single": 0.21, "esm_1v_ensemble": 0.245, "esm2_8m": 0.336, "esm2_35m": 0.367, "esm2_150m": 0.335, "esm2_650m": 0.271, "esm2_3b": 0.231, "esm2_15b": 0.21, "wavenet": 0.331, "rita_s": 0.266, "rita_m": 0.333, "rita_l": 0.372, "rita_xl": 0.384, "progen2_s": 0.279, "progen2_m": 0.347, "progen2_base": 0.402, "progen2_l": 0.345, "progen2_xl": 0.47, "gemme": 0.357, "vespa": 0.21, "vespal": 0.167, "vespag": 0.308, "protgpt2": 0.208, "tranception_s_no_retrieval": 0.198, "tranception_m_no_retrieval": 0.325, "tranception_l_no_retrieval": 0.398, "tranception_s": 0.275, "tranception_m": 0.368, "tranception_l": 0.436, "trancepteve_s": 0.334, "trancepteve_m": 0.37, "trancepteve_l": 0.388, "carp_38m": 0.237, "carp_600k": 0.335, "carp_640m": 0.058, "carp_76m": 0.243, "mif": 0.334, "mif_st": 0.327, "esm_if1": 0.377, "proteinmpnn": 0.326, "protssn_k_10_h_512": 0.264, "protssn_k_10_h_768": 0.245, "protssn_k_10_h_1280": 0.294, "protssn_k_20_h_512": 0.279, "protssn_k_20_h_768": 0.31, "protssn_k_20_h_1280": 0.281, "protssn_k_30_h_512": 0.298, "protssn_k_30_h_768": 0.28, "protssn_k_30_h_1280": 0.26, "protssn_ensemble": 0.287, "saprot_650m": 0.339, "saprot_35m": 0.331, "poet_200m": 0.302, "mulan": 0.32, "prosst_k_20": 0.455, "prosst_k_128": 0.41, "prosst_k_512": 0.386, "prosst_k_1024": 0.399, "prosst_k_2048": 0.372, "prosst_k_4096": 0.377, "escott": 0.328, "venusrem": 0.362, "rsalor": 0.272, "s2f": 0.3, "s2f_msa": 0.344, "s3f": 0.273, "s3f_msa": 0.326, "siterm": 0.324, "esm3_open_1_4b": 0.31, "esm_c_300m": 0.17, "esm_c_600m": 0.191, "xtrimopglm_1b_mlm": 0.241, "xtrimopglm_3b_mlm": 0.283, "xtrimopglm_10b_mlm": 0.235, "xtrimopglm_1b_clm": 0.323, "xtrimopglm_3b_clm": 0.282, "xtrimopglm_7b_clm": 0.261, "xtrimopglm_100b_int4": 0.268, "progen3_112m": 0.189, "progen3_219m": 0.203, "progen3_339m": 0.264, "progen3_762m": 0.235, "progen3_1b": 0.35, "progen3_3b": 0.352, "aido_protein_rag_16b": 0.444, "protriever": 0.281}}, {"id": "RASK_HUMAN_Weng_2022_binding-DARPin_K55", "label": "RASK_HUMAN_Weng_2022_binding-DARPin_K55", "protein": "RASK_HUMAN", "mean_mutations": 1.8760101314678568, "n_variants": 24873, "n_graph_configs": 24873, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6280193236714976, "epistasis.sign": 0.321256038647343, "epistasis.reciprocal_sign": 0.050724637681159424, "diminishing_returns_index": -0.3969517664967236, "increasing_costs_index": 0.12208778399799328, "global_idiosyncratic_index": 0.28186427760438093, "r_s_ratio": null, "gamma": 0.8241433235388359, "gamma_star": 0.5662549781596253, "local_optima_ratio": 0.023157640815341936, "autocorrelation": 0.7169561893124555, "fdc": -0.29874985337385884, "evolvability_enhancing_fraction": 0.00012018405329304306, "global_optima_accessibility": 0.5072970691110843}, "publication": {"title": "The energetic and allosteric landscape for KRAS inhibition", "journal": "Nature", "year": 2024, "doi": "10.1038/s41586-023-06954-0", "url": "https://doi.org/10.1038/s41586-023-06954-0"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.419, "evmutation": 0.47, "deepsequence_single": 0.602, "deepsequence_ensemble": 0.617, "eve_single": 0.598, "eve_ensemble": 0.598, "unirep": 0.268, "unirep_evotuned": 0.35, "msa_transformer_single": 0.641, "msa_transformer_ensemble": 0.639, "esm_1b": 0.585, "esm_1v_single": 0.524, "esm_1v_ensemble": 0.566, "esm2_8m": 0.573, "esm2_35m": 0.605, "esm2_150m": 0.621, "esm2_650m": 0.652, "esm2_3b": 0.563, "esm2_15b": 0.446, "wavenet": 0.677, "rita_s": 0.568, "rita_m": 0.511, "rita_l": 0.549, "rita_xl": 0.458, "progen2_s": 0.537, "progen2_m": 0.534, "progen2_base": 0.479, "progen2_l": 0.396, "progen2_xl": 0.358, "gemme": 0.569, "vespa": 0.529, "vespal": 0.49, "vespag": 0.634, "protgpt2": 0.289, "tranception_s_no_retrieval": 0.515, "tranception_m_no_retrieval": 0.589, "tranception_l_no_retrieval": 0.499, "tranception_s": 0.555, "tranception_m": 0.601, "tranception_l": 0.543, "trancepteve_s": 0.605, "trancepteve_m": 0.616, "trancepteve_l": 0.597, "carp_38m": 0.56, "carp_600k": 0.298, "carp_640m": 0.464, "carp_76m": 0.597, "mif": 0.239, "mif_st": 0.269, "esm_if1": 0.585, "proteinmpnn": 0.281, "protssn_k_10_h_512": 0.545, "protssn_k_10_h_768": 0.549, "protssn_k_10_h_1280": 0.542, "protssn_k_20_h_512": 0.552, "protssn_k_20_h_768": 0.556, "protssn_k_20_h_1280": 0.554, "protssn_k_30_h_512": 0.544, "protssn_k_30_h_768": 0.543, "protssn_k_30_h_1280": 0.51, "protssn_ensemble": 0.56, "saprot_650m": 0.601, "saprot_35m": 0.573, "poet_200m": 0.685, "mulan": 0.535, "prosst_k_20": 0.589, "prosst_k_128": 0.705, "prosst_k_512": 0.651, "prosst_k_1024": 0.696, "prosst_k_2048": 0.689, "prosst_k_4096": 0.71, "escott": 0.57, "venusrem": 0.671, "rsalor": 0.58, "s2f": 0.601, "s2f_msa": 0.627, "s3f": 0.637, "s3f_msa": 0.646, "siterm": 0.595, "esm3_open_1_4b": 0.584, "esm_c_300m": 0.645, "esm_c_600m": 0.602, "xtrimopglm_1b_mlm": 0.594, "xtrimopglm_3b_mlm": 0.564, "xtrimopglm_10b_mlm": 0.608, "xtrimopglm_1b_clm": 0.449, "xtrimopglm_3b_clm": 0.457, "xtrimopglm_7b_clm": 0.402, "xtrimopglm_100b_int4": 0.422, "progen3_112m": 0.573, "progen3_219m": 0.536, "progen3_339m": 0.593, "progen3_762m": 0.556, "progen3_1b": 0.563, "progen3_3b": 0.48, "aido_protein_rag_16b": 0.558, "protriever": 0.65}}, {"id": "SDA_BACSU_Tsuboyama_2023_1PV0", "label": "SDA_BACSU_Tsuboyama_2023_1PV0", "protein": "SDA_BACSU", "mean_mutations": 1.6989169675090252, "n_variants": 2770, "n_graph_configs": 2770, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6869295224507218, "epistasis.sign": 0.26775387530040523, "epistasis.reciprocal_sign": 0.045316602248872934, "diminishing_returns_index": -0.21726104620994882, "increasing_costs_index": 0.2812424710489894, "global_idiosyncratic_index": 0.2081835490337908, "r_s_ratio": 0.5993026361888151, "gamma": 0.8459735691610859, "gamma_star": 0.6416129202018489, "local_optima_ratio": 0.016245487364620937, "autocorrelation": 0.7497887589035841, "fdc": -0.6578974637556119, "evolvability_enhancing_fraction": 0.006887021136811581, "global_optima_accessibility": 0.0068592057761732855}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.631, "evmutation": 0.646, "deepsequence_single": 0.615, "deepsequence_ensemble": 0.619, "eve_single": 0.647, "eve_ensemble": 0.65, "unirep": 0.001, "unirep_evotuned": 0.303, "msa_transformer_single": 0.456, "msa_transformer_ensemble": 0.472, "esm_1b": 0.606, "esm_1v_single": 0.418, "esm_1v_ensemble": 0.466, "esm2_8m": 0.249, "esm2_35m": 0.274, "esm2_150m": 0.608, "esm2_650m": 0.594, "esm2_3b": 0.598, "esm2_15b": 0.596, "wavenet": 0.53, "rita_s": 0.026, "rita_m": 0.067, "rita_l": 0.003, "rita_xl": 0.238, "progen2_s": 0.081, "progen2_m": 0.123, "progen2_base": 0.045, "progen2_l": 0.119, "progen2_xl": 0.214, "gemme": 0.64, "vespa": 0.604, "vespal": 0.581, "vespag": 0.538, "protgpt2": -0.165, "tranception_s_no_retrieval": 0.01, "tranception_m_no_retrieval": 0.26, "tranception_l_no_retrieval": 0.107, "tranception_s": 0.607, "tranception_m": 0.613, "tranception_l": 0.368, "trancepteve_s": 0.652, "trancepteve_m": 0.65, "trancepteve_l": 0.486, "carp_38m": 0.037, "carp_600k": -0.079, "carp_640m": 0.336, "carp_76m": 0.011, "mif": 0.434, "mif_st": 0.41, "esm_if1": 0.608, "proteinmpnn": 0.597, "protssn_k_10_h_512": 0.65, "protssn_k_10_h_768": 0.641, "protssn_k_10_h_1280": 0.588, "protssn_k_20_h_512": 0.647, "protssn_k_20_h_768": 0.63, "protssn_k_20_h_1280": 0.613, "protssn_k_30_h_512": 0.622, "protssn_k_30_h_768": 0.623, "protssn_k_30_h_1280": 0.653, "protssn_ensemble": 0.633, "saprot_650m": 0.639, "saprot_35m": 0.636, "poet_200m": 0.546, "mulan": 0.542, "prosst_k_20": 0.706, "prosst_k_128": 0.535, "prosst_k_512": 0.703, "prosst_k_1024": 0.682, "prosst_k_2048": 0.726, "prosst_k_4096": 0.748, "escott": 0.641, "venusrem": 0.722, "rsalor": 0.641, "s2f": 0.611, "s2f_msa": 0.639, "s3f": 0.6, "s3f_msa": 0.638, "siterm": 0.584, "esm3_open_1_4b": 0.553, "esm_c_300m": 0.616, "esm_c_600m": 0.56, "xtrimopglm_1b_mlm": 0.472, "xtrimopglm_3b_mlm": 0.518, "xtrimopglm_10b_mlm": 0.548, "xtrimopglm_1b_clm": 0.27, "xtrimopglm_3b_clm": -0.022, "xtrimopglm_7b_clm": -0.041, "xtrimopglm_100b_int4": 0.6, "progen3_112m": 0.161, "progen3_219m": 0.15, "progen3_339m": 0.207, "progen3_762m": 0.094, "progen3_1b": 0.139, "progen3_3b": 0.226, "aido_protein_rag_16b": 0.61, "protriever": 0.582}}, {"id": "SPA_STAAU_Tsuboyama_2023_1LP1", "label": "SPA_STAAU_Tsuboyama_2023_1LP1", "protein": "SPA_STAAU", "mean_mutations": 1.508313539192399, "n_variants": 2105, "n_graph_configs": 2105, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5176511506579329, "epistasis.sign": 0.3531292871910807, "epistasis.reciprocal_sign": 0.12921956215098643, "diminishing_returns_index": -0.1649723419588069, "increasing_costs_index": 0.4454960934639111, "global_idiosyncratic_index": 0.2612718199790316, "r_s_ratio": 0.33007589725392555, "gamma": 0.672447723784598, "gamma_star": 0.3884315885069465, "local_optima_ratio": 0.02517814726840855, "autocorrelation": 0.6551322227944948, "fdc": -0.4713584317794711, "evolvability_enhancing_fraction": 0.011261778901401976, "global_optima_accessibility": 0.009026128266033254}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.432, "evmutation": 0.509, "deepsequence_single": 0.518, "deepsequence_ensemble": 0.501, "eve_single": 0.552, "eve_ensemble": 0.549, "unirep": -0.093, "unirep_evotuned": 0.495, "msa_transformer_single": 0.409, "msa_transformer_ensemble": 0.408, "esm_1b": 0.393, "esm_1v_single": -0.041, "esm_1v_ensemble": 0.001, "esm2_8m": -0.032, "esm2_35m": -0.033, "esm2_150m": -0.064, "esm2_650m": 0.005, "esm2_3b": -0.01, "esm2_15b": 0.02, "wavenet": 0.514, "rita_s": -0.154, "rita_m": 0.032, "rita_l": 0.205, "rita_xl": 0.105, "progen2_s": -0.081, "progen2_m": -0.041, "progen2_base": 0.057, "progen2_l": 0.338, "progen2_xl": 0.428, "gemme": 0.564, "vespa": 0.432, "vespal": 0.442, "vespag": 0.411, "protgpt2": -0.034, "tranception_s_no_retrieval": -0.011, "tranception_m_no_retrieval": 0.001, "tranception_l_no_retrieval": 0.066, "tranception_s": 0.437, "tranception_m": 0.43, "tranception_l": 0.427, "trancepteve_s": 0.548, "trancepteve_m": 0.547, "trancepteve_l": 0.536, "carp_38m": -0.137, "carp_600k": -0.087, "carp_640m": -0.071, "carp_76m": -0.009, "mif": 0.512, "mif_st": 0.453, "esm_if1": 0.705, "proteinmpnn": 0.56, "protssn_k_10_h_512": 0.52, "protssn_k_10_h_768": 0.511, "protssn_k_10_h_1280": 0.493, "protssn_k_20_h_512": 0.534, "protssn_k_20_h_768": 0.498, "protssn_k_20_h_1280": 0.515, "protssn_k_30_h_512": 0.51, "protssn_k_30_h_768": 0.501, "protssn_k_30_h_1280": 0.504, "protssn_ensemble": 0.513, "saprot_650m": 0.546, "saprot_35m": 0.414, "poet_200m": 0.495, "mulan": 0.424, "prosst_k_20": 0.328, "prosst_k_128": 0.524, "prosst_k_512": 0.515, "prosst_k_1024": 0.413, "prosst_k_2048": 0.617, "prosst_k_4096": 0.616, "escott": 0.609, "venusrem": 0.664, "rsalor": 0.662, "s2f": 0.422, "s2f_msa": 0.599, "s3f": 0.506, "s3f_msa": 0.628, "siterm": 0.528, "esm3_open_1_4b": 0.681, "esm_c_300m": -0.125, "esm_c_600m": 0.004, "xtrimopglm_1b_mlm": 0.098, "xtrimopglm_3b_mlm": 0.4, "xtrimopglm_10b_mlm": 0.124, "xtrimopglm_1b_clm": -0.072, "xtrimopglm_3b_clm": -0.015, "xtrimopglm_7b_clm": -0.235, "xtrimopglm_100b_int4": 0.427, "progen3_112m": -0.068, "progen3_219m": -0.016, "progen3_339m": 0.042, "progen3_762m": -0.031, "progen3_1b": 0.051, "progen3_3b": 0.067, "aido_protein_rag_16b": 0.702, "protriever": 0.528}}, {"id": "SPG1_STRSG_Olson_2014", "label": "SPG1_STRSG_Olson_2014", "protein": "SPG1_STRSG", "mean_mutations": 1.9980538660091403, "n_variants": 536962, "n_graph_configs": 536962, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6355832289440719, "epistasis.sign": 0.291061823004937, "epistasis.reciprocal_sign": 0.07335494805099109, "diminishing_returns_index": -0.48653949957613757, "increasing_costs_index": 0.08616843179189143, "global_idiosyncratic_index": 0.21799685238641842, "r_s_ratio": 1.2181246087197002, "gamma": 0.8255437572010116, "gamma_star": 0.5607769602081544, "local_optima_ratio": 0.0027934937667842418, "autocorrelation": 0.6955739000625822, "fdc": -0.18394978543158133, "evolvability_enhancing_fraction": 0.018745833925018156, "global_optima_accessibility": 0.9545070228433297}, "publication": {"title": "A comprehensive biophysical description of pairwise epistasis throughout an entire protein domain", "journal": "Current Biology", "year": 2014, "doi": "10.1016/j.cub.2014.09.072", "url": "https://doi.org/10.1016/j.cub.2014.09.072"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.239, "evmutation": 0.282, "deepsequence_single": -0.004, "deepsequence_ensemble": 0.006, "eve_single": 0.247, "eve_ensemble": 0.272, "unirep": -0.041, "unirep_evotuned": 0.017, "msa_transformer_single": -0.038, "msa_transformer_ensemble": 0.171, "esm_1b": 0.305, "esm_1v_single": 0.237, "esm_1v_ensemble": 0.192, "esm2_8m": 0.28, "esm2_35m": 0.232, "esm2_150m": 0.264, "esm2_650m": 0.301, "esm2_3b": 0.251, "esm2_15b": 0.337, "wavenet": 0.278, "rita_s": 0.252, "rita_m": 0.216, "rita_l": 0.214, "rita_xl": 0.208, "progen2_s": 0.232, "progen2_m": 0.222, "progen2_base": 0.218, "progen2_l": 0.239, "progen2_xl": 0.354, "gemme": 0.283, "vespa": 0.478, "vespal": 0.494, "vespag": 0.335, "protgpt2": 0.111, "tranception_s_no_retrieval": 0.223, "tranception_m_no_retrieval": 0.175, "tranception_l_no_retrieval": 0.279, "tranception_s": 0.243, "tranception_m": 0.204, "tranception_l": 0.289, "trancepteve_s": 0.229, "trancepteve_m": 0.188, "trancepteve_l": 0.29, "carp_38m": -0.092, "carp_600k": -0.112, "carp_640m": 0.158, "carp_76m": 0.004, "mif": 0.362, "mif_st": 0.306, "esm_if1": 0.423, "proteinmpnn": 0.147, "protssn_k_10_h_512": 0.38, "protssn_k_10_h_768": 0.315, "protssn_k_10_h_1280": 0.364, "protssn_k_20_h_512": 0.414, "protssn_k_20_h_768": 0.414, "protssn_k_20_h_1280": 0.36, "protssn_k_30_h_512": 0.418, "protssn_k_30_h_768": 0.417, "protssn_k_30_h_1280": 0.41, "protssn_ensemble": 0.407, "saprot_650m": 0.387, "saprot_35m": 0.374, "poet_200m": 0.36, "mulan": 0.336, "prosst_k_20": 0.736, "prosst_k_128": 0.776, "prosst_k_512": 0.787, "prosst_k_1024": 0.778, "prosst_k_2048": 0.743, "prosst_k_4096": 0.773, "escott": 0.402, "venusrem": 0.722, "rsalor": 0.408, "s2f": 0.388, "s2f_msa": 0.405, "s3f": 0.443, "s3f_msa": 0.437, "siterm": 0.243, "esm3_open_1_4b": 0.412, "esm_c_300m": 0.302, "esm_c_600m": 0.296, "xtrimopglm_1b_mlm": 0.257, "xtrimopglm_3b_mlm": 0.313, "xtrimopglm_10b_mlm": 0.32, "xtrimopglm_1b_clm": 0.215, "xtrimopglm_3b_clm": -0.056, "xtrimopglm_7b_clm": 0.115, "xtrimopglm_100b_int4": 0.341, "progen3_112m": 0.205, "progen3_219m": 0.181, "progen3_339m": 0.198, "progen3_762m": 0.192, "progen3_1b": 0.234, "progen3_3b": 0.266, "aido_protein_rag_16b": 0.538, "protriever": 0.291}}, {"id": "SPG1_STRSG_Wu_2016", "label": "SPG1_STRSG_Wu_2016", "protein": "SPG1_STRSG", "mean_mutations": 3.796270755222282, "n_variants": 149360, "n_graph_configs": 149360, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.435282871295733, "epistasis.sign": 0.34690188580863823, "epistasis.reciprocal_sign": 0.21781524289562876, "diminishing_returns_index": 0.27677664296549415, "increasing_costs_index": 0.9228567075420244, "global_idiosyncratic_index": 0.608493739227365, "r_s_ratio": 2.4771766394207093, "gamma": 0.45086307763225136, "gamma_star": 0.2549558460851446, "local_optima_ratio": 0.0012319228709159078, "autocorrelation": 0.3650345681715592, "fdc": -0.12308064858839449, "evolvability_enhancing_fraction": 0.00954989769202158, "global_optima_accessibility": 0.9975629351901446}, "publication": {"title": "Adaptation in protein fitness landscapes is facilitated by indirect paths", "journal": "eLife", "year": 2016, "doi": "10.7554/eLife.16965", "url": "https://doi.org/10.7554/eLife.16965"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": -0.051, "evmutation": 0.025, "deepsequence_single": 0.048, "deepsequence_ensemble": 0.048, "eve_single": 0.036, "eve_ensemble": 0.042, "unirep": 0.036, "unirep_evotuned": -0.048, "msa_transformer_single": 0.17, "msa_transformer_ensemble": 0.171, "esm_1b": 0.121, "esm_1v_single": 0.086, "esm_1v_ensemble": 0.059, "esm2_8m": 0.03, "esm2_35m": 0.097, "esm2_150m": 0.118, "esm2_650m": 0.128, "esm2_3b": 0.165, "esm2_15b": 0.199, "wavenet": 0.083, "rita_s": -0.012, "rita_m": -0.033, "rita_l": -0.051, "rita_xl": -0.023, "progen2_s": -0.061, "progen2_m": 0.016, "progen2_base": -0.014, "progen2_l": -0.031, "progen2_xl": -0.031, "gemme": 0.102, "vespa": 0.146, "vespal": 0.125, "vespag": 0.223, "protgpt2": -0.083, "tranception_s_no_retrieval": 0.001, "tranception_m_no_retrieval": -0.085, "tranception_l_no_retrieval": -0.024, "tranception_s": 0.039, "tranception_m": 0.013, "tranception_l": 0.021, "trancepteve_s": 0.044, "trancepteve_m": 0.027, "trancepteve_l": 0.027, "carp_38m": -0.013, "carp_600k": -0.099, "carp_640m": 0.04, "carp_76m": -0.026, "mif": 0.091, "mif_st": 0.023, "esm_if1": 0.18, "proteinmpnn": 0.089, "protssn_k_10_h_512": 0.144, "protssn_k_10_h_768": 0.136, "protssn_k_10_h_1280": 0.132, "protssn_k_20_h_512": 0.176, "protssn_k_20_h_768": 0.156, "protssn_k_20_h_1280": 0.131, "protssn_k_30_h_512": 0.194, "protssn_k_30_h_768": 0.167, "protssn_k_30_h_1280": 0.177, "protssn_ensemble": 0.159, "saprot_650m": 0.171, "saprot_35m": 0.114, "poet_200m": 0.114, "mulan": 0.154, "prosst_k_20": 0.3, "prosst_k_128": 0.283, "prosst_k_512": 0.297, "prosst_k_1024": 0.343, "prosst_k_2048": 0.247, "prosst_k_4096": 0.31, "escott": 0.121, "venusrem": 0.239, "rsalor": 0.149, "s2f": 0.161, "s2f_msa": 0.114, "s3f": 0.166, "s3f_msa": 0.118, "siterm": 0.037, "esm3_open_1_4b": 0.144, "esm_c_300m": 0.138, "esm_c_600m": 0.097, "xtrimopglm_1b_mlm": 0.091, "xtrimopglm_3b_mlm": 0.117, "xtrimopglm_10b_mlm": 0.117, "xtrimopglm_1b_clm": 0.011, "xtrimopglm_3b_clm": -0.001, "xtrimopglm_7b_clm": -0.039, "xtrimopglm_100b_int4": 0.072, "progen3_112m": -0.028, "progen3_219m": -0.007, "progen3_339m": -0.038, "progen3_762m": -0.037, "progen3_1b": 0.043, "progen3_3b": 0.023, "aido_protein_rag_16b": 0.347, "protriever": 0.16}}, {"id": "SPTN1_CHICK_Tsuboyama_2023_1TUD", "label": "SPTN1_CHICK_Tsuboyama_2023_1TUD", "protein": "SPTN1_CHICK", "mean_mutations": 1.6716651046547955, "n_variants": 3201, "n_graph_configs": 3201, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.7119991782626297, "epistasis.sign": 0.2525996507616176, "epistasis.reciprocal_sign": 0.035401170975752703, "diminishing_returns_index": -0.2541949911058981, "increasing_costs_index": 0.3409710157684005, "global_idiosyncratic_index": 0.2033688742935434, "r_s_ratio": 0.3587417937372454, "gamma": 0.8395736092230328, "gamma_star": 0.676598007286877, "local_optima_ratio": 0.01811933770696657, "autocorrelation": 0.6384675102062362, "fdc": -0.48206020787534065, "evolvability_enhancing_fraction": 0.006346893084949098, "global_optima_accessibility": 0.005935645110902843}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.572, "evmutation": 0.613, "deepsequence_single": 0.647, "deepsequence_ensemble": 0.599, "eve_single": 0.632, "eve_ensemble": 0.615, "unirep": 0.23, "unirep_evotuned": 0.524, "msa_transformer_single": 0.552, "msa_transformer_ensemble": 0.593, "esm_1b": 0.637, "esm_1v_single": 0.649, "esm_1v_ensemble": 0.641, "esm2_8m": -0.133, "esm2_35m": 0.619, "esm2_150m": 0.627, "esm2_650m": 0.637, "esm2_3b": 0.726, "esm2_15b": 0.657, "wavenet": 0.624, "rita_s": 0.578, "rita_m": 0.643, "rita_l": 0.604, "rita_xl": 0.589, "progen2_s": 0.589, "progen2_m": 0.613, "progen2_base": 0.652, "progen2_l": 0.563, "progen2_xl": 0.601, "gemme": 0.621, "vespa": 0.631, "vespal": 0.632, "vespag": 0.683, "protgpt2": 0.457, "tranception_s_no_retrieval": 0.508, "tranception_m_no_retrieval": 0.57, "tranception_l_no_retrieval": 0.548, "tranception_s": 0.629, "tranception_m": 0.641, "tranception_l": 0.621, "trancepteve_s": 0.623, "trancepteve_m": 0.628, "trancepteve_l": 0.618, "carp_38m": 0.488, "carp_600k": -0.221, "carp_640m": 0.465, "carp_76m": 0.498, "mif": 0.421, "mif_st": 0.401, "esm_if1": 0.62, "proteinmpnn": 0.623, "protssn_k_10_h_512": 0.693, "protssn_k_10_h_768": 0.679, "protssn_k_10_h_1280": 0.683, "protssn_k_20_h_512": 0.685, "protssn_k_20_h_768": 0.696, "protssn_k_20_h_1280": 0.685, "protssn_k_30_h_512": 0.676, "protssn_k_30_h_768": 0.674, "protssn_k_30_h_1280": 0.696, "protssn_ensemble": 0.691, "saprot_650m": 0.667, "saprot_35m": 0.526, "poet_200m": 0.667, "mulan": 0.211, "prosst_k_20": 0.587, "prosst_k_128": 0.546, "prosst_k_512": 0.576, "prosst_k_1024": 0.704, "prosst_k_2048": 0.762, "prosst_k_4096": 0.744, "escott": 0.701, "venusrem": 0.77, "rsalor": 0.662, "s2f": 0.656, "s2f_msa": 0.662, "s3f": 0.691, "s3f_msa": 0.673, "siterm": 0.686, "esm3_open_1_4b": 0.62, "esm_c_300m": 0.65, "esm_c_600m": 0.664, "xtrimopglm_1b_mlm": 0.546, "xtrimopglm_3b_mlm": 0.576, "xtrimopglm_10b_mlm": 0.623, "xtrimopglm_1b_clm": 0.562, "xtrimopglm_3b_clm": 0.352, "xtrimopglm_7b_clm": 0.413, "xtrimopglm_100b_int4": 0.514, "progen3_112m": 0.302, "progen3_219m": 0.258, "progen3_339m": 0.459, "progen3_762m": 0.416, "progen3_1b": 0.543, "progen3_3b": 0.568, "aido_protein_rag_16b": 0.708, "protriever": 0.595}}, {"id": "UBE4B_HUMAN_Tsuboyama_2023_3L1X", "label": "UBE4B_HUMAN_Tsuboyama_2023_3L1X", "protein": "UBE4B_HUMAN", "mean_mutations": 1.6913307564881281, "n_variants": 3622, "n_graph_configs": 3622, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6087481347583982, "epistasis.sign": 0.31840129080784674, "epistasis.reciprocal_sign": 0.07285057443375512, "diminishing_returns_index": -0.3319459092445111, "increasing_costs_index": 0.37178406359651894, "global_idiosyncratic_index": 0.38055892120349566, "r_s_ratio": 0.5184618807619956, "gamma": 0.6491403245004037, "gamma_star": 0.535897560324643, "local_optima_ratio": 0.02236333517393705, "autocorrelation": 0.5305459085609912, "fdc": -0.1696261091252991, "evolvability_enhancing_fraction": 0.005922459120587046, "global_optima_accessibility": 0.127829928216455}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.325, "evmutation": 0.443, "deepsequence_single": 0.464, "deepsequence_ensemble": 0.467, "eve_single": 0.466, "eve_ensemble": 0.466, "unirep": 0.051, "unirep_evotuned": 0.346, "msa_transformer_single": 0.379, "msa_transformer_ensemble": 0.415, "esm_1b": 0.391, "esm_1v_single": 0.453, "esm_1v_ensemble": 0.449, "esm2_8m": 0.347, "esm2_35m": 0.42, "esm2_150m": 0.49, "esm2_650m": 0.46, "esm2_3b": 0.467, "esm2_15b": 0.477, "wavenet": 0.421, "rita_s": 0.142, "rita_m": 0.353, "rita_l": 0.357, "rita_xl": 0.364, "progen2_s": 0.345, "progen2_m": 0.374, "progen2_base": 0.343, "progen2_l": 0.322, "progen2_xl": 0.355, "gemme": 0.486, "vespa": 0.475, "vespal": 0.453, "vespag": 0.448, "protgpt2": 0.441, "tranception_s_no_retrieval": 0.079, "tranception_m_no_retrieval": 0.11, "tranception_l_no_retrieval": 0.255, "tranception_s": 0.383, "tranception_m": 0.385, "tranception_l": 0.329, "trancepteve_s": 0.472, "trancepteve_m": 0.477, "trancepteve_l": 0.424, "carp_38m": 0.359, "carp_600k": -0.095, "carp_640m": 0.348, "carp_76m": 0.344, "mif": 0.488, "mif_st": 0.409, "esm_if1": 0.52, "proteinmpnn": 0.351, "protssn_k_10_h_512": 0.411, "protssn_k_10_h_768": 0.436, "protssn_k_10_h_1280": 0.424, "protssn_k_20_h_512": 0.447, "protssn_k_20_h_768": 0.448, "protssn_k_20_h_1280": 0.428, "protssn_k_30_h_512": 0.407, "protssn_k_30_h_768": 0.429, "protssn_k_30_h_1280": 0.42, "protssn_ensemble": 0.432, "saprot_650m": 0.5, "saprot_35m": 0.407, "poet_200m": 0.441, "mulan": 0.321, "prosst_k_20": 0.454, "prosst_k_128": 0.421, "prosst_k_512": 0.437, "prosst_k_1024": 0.5, "prosst_k_2048": 0.522, "prosst_k_4096": 0.468, "escott": 0.527, "venusrem": 0.515, "rsalor": 0.496, "s2f": 0.418, "s2f_msa": 0.454, "s3f": 0.458, "s3f_msa": 0.472, "siterm": 0.425, "esm3_open_1_4b": 0.558, "esm_c_300m": 0.488, "esm_c_600m": 0.511, "xtrimopglm_1b_mlm": 0.421, "xtrimopglm_3b_mlm": 0.424, "xtrimopglm_10b_mlm": 0.391, "xtrimopglm_1b_clm": 0.231, "xtrimopglm_3b_clm": -0.034, "xtrimopglm_7b_clm": -0.16, "xtrimopglm_100b_int4": 0.342, "progen3_112m": -0.066, "progen3_219m": -0.065, "progen3_339m": 0.29, "progen3_762m": 0.332, "progen3_1b": 0.418, "progen3_3b": 0.389, "aido_protein_rag_16b": 0.563, "protriever": 0.434}}, {"id": "VILI_CHICK_Tsuboyama_2023_1YU5", "label": "VILI_CHICK_Tsuboyama_2023_1YU5", "protein": "VILI_CHICK", "mean_mutations": 1.5319314641744548, "n_variants": 2568, "n_graph_configs": 2568, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6587077969395191, "epistasis.sign": 0.2946320136021375, "epistasis.reciprocal_sign": 0.04666018945834345, "diminishing_returns_index": -0.37986402357747373, "increasing_costs_index": 0.11073455712282773, "global_idiosyncratic_index": 0.2080482483114451, "r_s_ratio": 0.40772304842770285, "gamma": 0.7964559643305905, "gamma_star": 0.6120476074811756, "local_optima_ratio": 0.02764797507788162, "autocorrelation": 0.727248679836531, "fdc": -0.620819704521917, "evolvability_enhancing_fraction": 0.008185852681712399, "global_optima_accessibility": 0.007398753894080997}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.284, "evmutation": 0.422, "deepsequence_single": 0.594, "deepsequence_ensemble": 0.63, "eve_single": 0.634, "eve_ensemble": 0.637, "unirep": 0.162, "unirep_evotuned": 0.54, "msa_transformer_single": 0.553, "msa_transformer_ensemble": 0.59, "esm_1b": 0.673, "esm_1v_single": 0.63, "esm_1v_ensemble": 0.673, "esm2_8m": 0.371, "esm2_35m": 0.316, "esm2_150m": 0.652, "esm2_650m": 0.662, "esm2_3b": 0.589, "esm2_15b": 0.478, "wavenet": 0.592, "rita_s": 0.293, "rita_m": 0.431, "rita_l": 0.601, "rita_xl": 0.524, "progen2_s": 0.424, "progen2_m": 0.575, "progen2_base": 0.494, "progen2_l": 0.482, "progen2_xl": 0.579, "gemme": 0.599, "vespa": 0.701, "vespal": 0.694, "vespag": 0.665, "protgpt2": 0.322, "tranception_s_no_retrieval": 0.346, "tranception_m_no_retrieval": 0.515, "tranception_l_no_retrieval": 0.521, "tranception_s": 0.487, "tranception_m": 0.564, "tranception_l": 0.563, "trancepteve_s": 0.634, "trancepteve_m": 0.659, "trancepteve_l": 0.642, "carp_38m": 0.272, "carp_600k": 0.283, "carp_640m": 0.426, "carp_76m": 0.421, "mif": 0.542, "mif_st": 0.433, "esm_if1": 0.775, "proteinmpnn": 0.702, "protssn_k_10_h_512": 0.672, "protssn_k_10_h_768": 0.658, "protssn_k_10_h_1280": 0.666, "protssn_k_20_h_512": 0.658, "protssn_k_20_h_768": 0.653, "protssn_k_20_h_1280": 0.665, "protssn_k_30_h_512": 0.649, "protssn_k_30_h_768": 0.656, "protssn_k_30_h_1280": 0.652, "protssn_ensemble": 0.662, "saprot_650m": 0.685, "saprot_35m": 0.798, "poet_200m": 0.644, "mulan": 0.449, "prosst_k_20": 0.601, "prosst_k_128": 0.742, "prosst_k_512": 0.798, "prosst_k_1024": 0.805, "prosst_k_2048": 0.816, "prosst_k_4096": 0.83, "escott": 0.593, "venusrem": 0.799, "rsalor": 0.677, "s2f": 0.61, "s2f_msa": 0.644, "s3f": 0.65, "s3f_msa": 0.664, "siterm": 0.534, "esm3_open_1_4b": 0.781, "esm_c_300m": 0.659, "esm_c_600m": 0.648, "xtrimopglm_1b_mlm": 0.625, "xtrimopglm_3b_mlm": 0.618, "xtrimopglm_10b_mlm": 0.633, "xtrimopglm_1b_clm": 0.696, "xtrimopglm_3b_clm": 0.518, "xtrimopglm_7b_clm": 0.461, "xtrimopglm_100b_int4": 0.564, "progen3_112m": 0.162, "progen3_219m": 0.141, "progen3_339m": 0.318, "progen3_762m": 0.407, "progen3_1b": 0.423, "progen3_3b": 0.604, "aido_protein_rag_16b": 0.724, "protriever": 0.64}}, {"id": "YAIA_ECOLI_Tsuboyama_2023_2KVT", "label": "YAIA_ECOLI_Tsuboyama_2023_2KVT", "protein": "YAIA_ECOLI", "mean_mutations": 1.5089947089947089, "n_variants": 1890, "n_graph_configs": 1890, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.6392106327824145, "epistasis.sign": 0.3021861935657881, "epistasis.reciprocal_sign": 0.058603173651797354, "diminishing_returns_index": -0.4645248932180338, "increasing_costs_index": 0.1355165366094544, "global_idiosyncratic_index": 0.3436284765578297, "r_s_ratio": 0.5243177576619953, "gamma": 0.732113481166608, "gamma_star": 0.5806074591306172, "local_optima_ratio": 0.025925925925925925, "autocorrelation": 0.6341582873352728, "fdc": -0.5890986709304986, "evolvability_enhancing_fraction": 0.0038005721291377195, "global_optima_accessibility": 0.010052910052910053}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.291, "evmutation": 0.602, "deepsequence_single": 0.572, "deepsequence_ensemble": 0.596, "eve_single": 0.599, "eve_ensemble": 0.591, "unirep": -0.221, "unirep_evotuned": 0.533, "msa_transformer_single": 0.559, "msa_transformer_ensemble": 0.598, "esm_1b": 0.593, "esm_1v_single": 0.27, "esm_1v_ensemble": 0.481, "esm2_8m": -0.107, "esm2_35m": 0.173, "esm2_150m": 0.572, "esm2_650m": 0.625, "esm2_3b": 0.684, "esm2_15b": 0.684, "wavenet": 0.591, "rita_s": -0.091, "rita_m": -0.182, "rita_l": -0.13, "rita_xl": 0.264, "progen2_s": -0.125, "progen2_m": -0.166, "progen2_base": -0.231, "progen2_l": -0.023, "progen2_xl": 0.52, "gemme": 0.605, "vespa": 0.681, "vespal": 0.643, "vespag": 0.678, "protgpt2": -0.06, "tranception_s_no_retrieval": -0.234, "tranception_m_no_retrieval": -0.064, "tranception_l_no_retrieval": 0.502, "tranception_s": 0.349, "tranception_m": 0.359, "tranception_l": 0.528, "trancepteve_s": 0.572, "trancepteve_m": 0.569, "trancepteve_l": 0.604, "carp_38m": -0.046, "carp_600k": -0.261, "carp_640m": 0.208, "carp_76m": -0.013, "mif": 0.449, "mif_st": 0.425, "esm_if1": 0.651, "proteinmpnn": 0.544, "protssn_k_10_h_512": 0.587, "protssn_k_10_h_768": 0.627, "protssn_k_10_h_1280": 0.585, "protssn_k_20_h_512": 0.619, "protssn_k_20_h_768": 0.612, "protssn_k_20_h_1280": 0.627, "protssn_k_30_h_512": 0.625, "protssn_k_30_h_768": 0.657, "protssn_k_30_h_1280": 0.677, "protssn_ensemble": 0.632, "saprot_650m": 0.616, "saprot_35m": 0.498, "poet_200m": 0.588, "mulan": -0.035, "prosst_k_20": 0.607, "prosst_k_128": 0.513, "prosst_k_512": 0.526, "prosst_k_1024": 0.619, "prosst_k_2048": 0.792, "prosst_k_4096": 0.743, "escott": 0.662, "venusrem": 0.753, "rsalor": 0.709, "s2f": 0.648, "s2f_msa": 0.657, "s3f": 0.668, "s3f_msa": 0.671, "siterm": 0.441, "esm3_open_1_4b": 0.752, "esm_c_300m": 0.229, "esm_c_600m": 0.39, "xtrimopglm_1b_mlm": 0.106, "xtrimopglm_3b_mlm": 0.162, "xtrimopglm_10b_mlm": 0.505, "xtrimopglm_1b_clm": -0.052, "xtrimopglm_3b_clm": -0.222, "xtrimopglm_7b_clm": -0.048, "xtrimopglm_100b_int4": 0.637, "progen3_112m": -0.074, "progen3_219m": 0.059, "progen3_339m": 0.358, "progen3_762m": 0.314, "progen3_1b": 0.395, "progen3_3b": 0.496, "aido_protein_rag_16b": 0.744, "protriever": 0.634}}, {"id": "YAP1_HUMAN_Araya_2012", "label": "YAP1_HUMAN_Araya_2012", "protein": "YAP1_HUMAN", "mean_mutations": 1.9640694789081885, "n_variants": 10075, "n_graph_configs": 10053, "n_isolated_removed": 22, "features": {"epistasis.magnitude": 0.5179134701507068, "epistasis.sign": 0.3552319015537988, "epistasis.reciprocal_sign": 0.12685462829549438, "diminishing_returns_index": 0.1418400814427177, "increasing_costs_index": 0.8100023449909992, "global_idiosyncratic_index": 0.9296465036138222, "r_s_ratio": 0.9117337070555028, "gamma": 0.46700932748482293, "gamma_star": 0.39105884185521245, "local_optima_ratio": 0.09211180742067045, "autocorrelation": 0.3691232231877749, "fdc": -0.24293488813720707, "evolvability_enhancing_fraction": 0.0, "global_optima_accessibility": 0.7243608872973242}, "publication": {"title": "A fundamental protein property, thermodynamic stability, revealed solely from large-scale measurements of protein function", "journal": "Proceedings of the National Academy of Sciences", "year": 2012, "doi": "10.1073/pnas.1209751109", "url": "https://doi.org/10.1073/pnas.1209751109"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.428, "evmutation": 0.321, "deepsequence_single": 0.464, "deepsequence_ensemble": 0.46, "eve_single": 0.449, "eve_ensemble": 0.455, "unirep": 0.329, "unirep_evotuned": 0.289, "msa_transformer_single": 0.018, "msa_transformer_ensemble": 0.036, "esm_1b": 0.333, "esm_1v_single": 0.28, "esm_1v_ensemble": 0.285, "esm2_8m": 0.411, "esm2_35m": 0.457, "esm2_150m": 0.451, "esm2_650m": 0.466, "esm2_3b": 0.382, "esm2_15b": 0.313, "wavenet": 0.189, "rita_s": 0.18, "rita_m": 0.176, "rita_l": 0.158, "rita_xl": 0.167, "progen2_s": 0.299, "progen2_m": 0.226, "progen2_base": 0.257, "progen2_l": 0.207, "progen2_xl": 0.15, "gemme": 0.332, "vespa": 0.425, "vespal": 0.451, "vespag": 0.35, "protgpt2": 0.139, "tranception_s_no_retrieval": 0.302, "tranception_m_no_retrieval": 0.196, "tranception_l_no_retrieval": 0.217, "tranception_s": 0.402, "tranception_m": 0.326, "tranception_l": 0.359, "trancepteve_s": 0.435, "trancepteve_m": 0.405, "trancepteve_l": 0.439, "carp_38m": 0.485, "carp_600k": -0.104, "carp_640m": 0.335, "carp_76m": 0.486, "mif": 0.34, "mif_st": 0.362, "esm_if1": 0.383, "proteinmpnn": 0.197, "protssn_k_10_h_512": 0.446, "protssn_k_10_h_768": 0.379, "protssn_k_10_h_1280": 0.432, "protssn_k_20_h_512": 0.426, "protssn_k_20_h_768": 0.417, "protssn_k_20_h_1280": 0.412, "protssn_k_30_h_512": 0.415, "protssn_k_30_h_768": 0.441, "protssn_k_30_h_1280": 0.444, "protssn_ensemble": 0.438, "saprot_650m": 0.386, "saprot_35m": 0.422, "poet_200m": 0.421, "mulan": 0.408, "prosst_k_20": 0.586, "prosst_k_128": 0.579, "prosst_k_512": 0.58, "prosst_k_1024": 0.575, "prosst_k_2048": 0.549, "prosst_k_4096": 0.59, "escott": 0.392, "venusrem": 0.546, "rsalor": 0.519, "s2f": 0.458, "s2f_msa": 0.493, "s3f": 0.459, "s3f_msa": 0.496, "siterm": 0.376, "esm3_open_1_4b": 0.439, "esm_c_300m": 0.396, "esm_c_600m": 0.298, "xtrimopglm_1b_mlm": 0.253, "xtrimopglm_3b_mlm": 0.166, "xtrimopglm_10b_mlm": 0.102, "xtrimopglm_1b_clm": 0.296, "xtrimopglm_3b_clm": 0.351, "xtrimopglm_7b_clm": 0.144, "xtrimopglm_100b_int4": 0.061, "progen3_112m": 0.384, "progen3_219m": 0.322, "progen3_339m": 0.284, "progen3_762m": 0.243, "progen3_1b": 0.247, "progen3_3b": 0.232, "aido_protein_rag_16b": 0.48, "protriever": 0.365}}, {"id": "YNZC_BACSU_Tsuboyama_2023_2JVD", "label": "YNZC_BACSU_Tsuboyama_2023_2JVD", "protein": "YNZC_BACSU", "mean_mutations": 1.6895652173913043, "n_variants": 2300, "n_graph_configs": 2300, "n_isolated_removed": 0, "features": {"epistasis.magnitude": 0.5620383248338742, "epistasis.sign": 0.3445217122546747, "epistasis.reciprocal_sign": 0.09343996291145108, "diminishing_returns_index": -0.2236319234356638, "increasing_costs_index": 0.32390882574465085, "global_idiosyncratic_index": 0.2819404475865268, "r_s_ratio": 0.5124799871799413, "gamma": 0.6916835865469493, "gamma_star": 0.4685983619224231, "local_optima_ratio": 0.017391304347826087, "autocorrelation": 0.7280399748628505, "fdc": -0.6317380932630248, "evolvability_enhancing_fraction": 0.010686829433133396, "global_optima_accessibility": 0.007391304347826087}, "publication": {"title": "Mega-scale experimental analysis of protein folding stability in biology and design", "journal": "Nature", "year": 2023, "doi": "10.1038/s41586-023-06328-6", "url": "https://doi.org/10.1038/s41586-023-06328-6"}, "source_url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "outcomes": {"site_independent": 0.596, "evmutation": 0.61, "deepsequence_single": 0.595, "deepsequence_ensemble": 0.618, "eve_single": 0.604, "eve_ensemble": 0.609, "unirep": 0.492, "unirep_evotuned": 0.394, "msa_transformer_single": 0.591, "msa_transformer_ensemble": 0.583, "esm_1b": 0.673, "esm_1v_single": 0.649, "esm_1v_ensemble": 0.667, "esm2_8m": 0.584, "esm2_35m": 0.621, "esm2_150m": 0.65, "esm2_650m": 0.658, "esm2_3b": 0.631, "esm2_15b": 0.63, "wavenet": 0.525, "rita_s": 0.488, "rita_m": 0.534, "rita_l": 0.612, "rita_xl": 0.604, "progen2_s": 0.594, "progen2_m": 0.611, "progen2_base": 0.587, "progen2_l": 0.59, "progen2_xl": 0.643, "gemme": 0.594, "vespa": 0.586, "vespal": 0.547, "vespag": 0.6, "protgpt2": 0.568, "tranception_s_no_retrieval": 0.573, "tranception_m_no_retrieval": 0.567, "tranception_l_no_retrieval": 0.544, "tranception_s": 0.665, "tranception_m": 0.634, "tranception_l": 0.621, "trancepteve_s": 0.669, "trancepteve_m": 0.624, "trancepteve_l": 0.63, "carp_38m": 0.446, "carp_600k": 0.353, "carp_640m": 0.536, "carp_76m": 0.448, "mif": 0.483, "mif_st": 0.562, "esm_if1": 0.68, "proteinmpnn": 0.65, "protssn_k_10_h_512": 0.701, "protssn_k_10_h_768": 0.649, "protssn_k_10_h_1280": 0.665, "protssn_k_20_h_512": 0.687, "protssn_k_20_h_768": 0.7, "protssn_k_20_h_1280": 0.691, "protssn_k_30_h_512": 0.687, "protssn_k_30_h_768": 0.676, "protssn_k_30_h_1280": 0.679, "protssn_ensemble": 0.69, "saprot_650m": 0.638, "saprot_35m": 0.76, "poet_200m": 0.591, "mulan": 0.39, "prosst_k_20": 0.729, "prosst_k_128": 0.717, "prosst_k_512": 0.739, "prosst_k_1024": 0.748, "prosst_k_2048": 0.719, "prosst_k_4096": 0.745, "escott": 0.592, "venusrem": 0.693, "rsalor": 0.618, "s2f": 0.693, "s2f_msa": 0.675, "s3f": 0.703, "s3f_msa": 0.68, "siterm": 0.578, "esm3_open_1_4b": 0.671, "esm_c_300m": 0.589, "esm_c_600m": 0.576, "xtrimopglm_1b_mlm": 0.655, "xtrimopglm_3b_mlm": 0.602, "xtrimopglm_10b_mlm": 0.642, "xtrimopglm_1b_clm": 0.591, "xtrimopglm_3b_clm": -0.09, "xtrimopglm_7b_clm": 0.109, "xtrimopglm_100b_int4": 0.605, "progen3_112m": 0.603, "progen3_219m": 0.592, "progen3_339m": 0.672, "progen3_762m": 0.584, "progen3_1b": 0.577, "progen3_3b": 0.609, "aido_protein_rag_16b": 0.631, "protriever": 0.61}}], "provenance": {"status": "completed_with_missing_metrics", "dataset": "ProteinGym DMS substitution benchmark", "dataset_release": "v1.3", "dataset_release_record": "https://zenodo.org/records/15293562", "reference_rows": 217, "assay_archive_files": 217, "assay_archive_variants": 2465767, "eligible_assays": 36, "eligibility_rule": "unfiltered mean number of substitution labels per measured variant > 1.5; indels are excluded by the substitution-only benchmark", "eligible_variants": 1788585, "selected_dms_ids": ["A4_HUMAN_Seuma_2022", "AMFR_HUMAN_Tsuboyama_2023_4G3O", "CAPSD_AAV2S_Sinai_2021", "CBX4_HUMAN_Tsuboyama_2023_2K28", "CSN4_MOUSE_Tsuboyama_2023_1UFM", "D7PM05_CLYGR_Somermeyer_2022", "DLG4_HUMAN_Faure_2021", "DOCK1_MOUSE_Tsuboyama_2023_2M0Y", "F7YBW8_MESOW_Aakre_2015", "F7YBW8_MESOW_Ding_2023", "GCN4_YEAST_Staller_2018", "GFP_AEQVI_Sarkisyan_2016", "GRB2_HUMAN_Faure_2021", "HECD1_HUMAN_Tsuboyama_2023_3DKM", "HIS7_YEAST_Pokusaeva_2019", "ISDH_STAAW_Tsuboyama_2023_2LHR", "MYO3_YEAST_Tsuboyama_2023_2BTT", "NKX31_HUMAN_Tsuboyama_2023_2L9R", "OBSCN_HUMAN_Tsuboyama_2023_1V1C", "PABP_YEAST_Melamed_2013", "PHOT_CHLRE_Chen_2023", "POLG_PESV_Tsuboyama_2023_2MXD", "Q6WV12_9MAXI_Somermeyer_2022", "Q8WTC7_9CNID_Somermeyer_2022", "RASK_HUMAN_Weng_2022_abundance", "RASK_HUMAN_Weng_2022_binding-DARPin_K55", "SDA_BACSU_Tsuboyama_2023_1PV0", "SPA_STAAU_Tsuboyama_2023_1LP1", "SPG1_STRSG_Olson_2014", "SPG1_STRSG_Wu_2016", "SPTN1_CHICK_Tsuboyama_2023_1TUD", "UBE4B_HUMAN_Tsuboyama_2023_3L1X", "VILI_CHICK_Tsuboyama_2023_1YU5", "YAIA_ECOLI_Tsuboyama_2023_2KVT", "YAP1_HUMAN_Araya_2012", "YNZC_BACSU_Tsuboyama_2023_2JVD"], "missing_published_scores": {"GFP_AEQVI_Sarkisyan_2016": {"Protriever": "The pinned official ProteinGym per-DMS Spearman table has no score in this cell."}, "GRB2_HUMAN_Faure_2021": {"Protriever": "The pinned official ProteinGym per-DMS Spearman table has no score in this cell."}, "HIS7_YEAST_Pokusaeva_2019": {"Protriever": "The pinned official ProteinGym per-DMS Spearman table has no score in this cell."}}, "source_files": {"DMS_ProteinGym_substitutions.zip": {"url": "https://zenodo.org/records/15293562/files/DMS_ProteinGym_substitutions.zip?download=1", "sha256": "3a83766254ac9ac9984ec25cb73c6e010ea4418f5e35f143933e6b6e6473b921"}, "DMS_substitutions.csv": {"url": "https://zenodo.org/records/15293562/files/DMS_substitutions.csv?download=1", "sha256": "a8f498011532a74aa9fe556a50555a75e928c5837d19c06a87592ae04049b308"}, "DMS_substitutions_Spearman_DMS_level.csv": {"url": "https://raw.githubusercontent.com/OATML-Markslab/ProteinGym/144fe22b07dfaeec2b366f2346203a9838a55b4c/benchmarks/DMS_zero_shot/substitutions/Spearman/DMS_substitutions_Spearman_DMS_level.csv", "repository_commit": "144fe22b07dfaeec2b366f2346203a9838a55b4c", "sha256": "f432423b87f79ac9778dfac86e3d95be041d246bc618dc0a406b35b0b7466437", "scope": "published per-DMS zero-shot Spearman correlations; no models were trained or scored here"}, "proteingym_dms_citations.json": {"path": "docs/home/insights/references/proteingym_dms_citations.json", "sha256": "24820f710360764ba402a698ddc0162a613cbb31baf075526a9e6341b8be307f", "scope": "verified per-assay peer-reviewed publication metadata"}}, "graphfla": {"repository_commit": "22cbd7ac89db01322aa6beb32196e201e6281ba7", "landscape": "ProteinLandscape over the union of observed mutated positions; DMS_score; maximize=True; n_edit=1; epsilon=0", "isolate_handling": "GraphFLA's standard construction removes configurations with no observed one-edit neighbors after eligibility and input validation; per-record n_graph_configs and n_isolated_removed are retained", "ee_equivalence_validation": {"path": "docs/home/insights/proteingym/ee_validation.json", "sha256": "1c806e60c84d4137d9fde8a0b04a5f0db051891b31bfd249c39cfc3f23b046ac", "status": "optimized helper matches public EE and prior exact helper on two full landscapes, including a neutral pair"}, "sparse_gamma_validation": {"path": "docs/home/insights/proteingym/gamma_validation.json", "sha256": "135959ba3f2bded18cab0dd292fab1d0df890ec891b7534698c41a840384dcef", "status": "bundled exact sparse traversal compared with GraphFLA's joint public API on full ProteinGym assays"}, "bounded_fdc_validation": {"path": "docs/home/insights/proteingym/fdc_validation.json", "sha256": "a109a09f6da016052ce858d1550b75d873a334d55f25e00789c72de06a8ad321", "status": "exact match with GraphFLA public FDC for single and tied global optima"}, "preparation_code": {"script": {"path": "docs/home/insights/proteingym/prepare_proteingym.py", "sha256": "42f06ba12b0a543154e07c2fd85c3118516479e73ba27c12a00697b7d08ef683"}, "sparse_gamma_helper": {"path": "docs/home/insights/proteingym/sparse_gamma.py", "sha256": "082e98fe5e40abd7e17ae726233bd7baec830d484f108b52c5dbecb3bbc3cc02"}, "bounded_ee_helper": {"path": "docs/home/insights/proteingym/bounded_ee_fast.py", "sha256": "210c979885f75a8ceada0f854cf52ad983a92f9d90d6383a4671772faac4706d"}}, "metric_failures": {"D7PM05_CLYGR_Somermeyer_2022": {"r_s_ratio": "UserWarning: The additive design is rank deficient; slope is not identifiable. Returning nan."}, "GCN4_YEAST_Staller_2018": {"r_s_ratio": "UserWarning: The additive coefficients are not identifiable. Returning nan."}, "Q8WTC7_9CNID_Somermeyer_2022": {"r_s_ratio": "UserWarning: The additive design is rank deficient; slope is not identifiable. Returning nan."}, "RASK_HUMAN_Weng_2022_abundance": {"r_s_ratio": "UserWarning: The encoded model exceeds the 64 MiB QR workspace limit. Returning nan."}, "RASK_HUMAN_Weng_2022_binding-DARPin_K55": {"r_s_ratio": "UserWarning: The encoded model exceeds the 64 MiB QR workspace limit. Returning nan."}}, "parameters_by_dms": {"A4_HUMAN_Seuma_2022": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1}, "gamma_star": {"n_jobs": 1}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "AMFR_HUMAN_Tsuboyama_2023_4G3O": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1}, "gamma_star": {"n_jobs": 1}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "CAPSD_AAV2S_Sinai_2021": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "CBX4_HUMAN_Tsuboyama_2023_2K28": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "CSN4_MOUSE_Tsuboyama_2023_1UFM": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "D7PM05_CLYGR_Somermeyer_2022": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 0.88868, "diagnostics": {"rows": 12138, "variables": 234, "sparse_mutations": 23946, "candidate_position_pairs": 7692, "complete_square_contexts": 7798}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 0.88868, "diagnostics": {"rows": 12138, "variables": 234, "sparse_mutations": 23946, "candidate_position_pairs": 7692, "complete_square_contexts": 7798}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "DLG4_HUMAN_Faure_2021": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1}, "gamma_star": {"n_jobs": 1}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "DOCK1_MOUSE_Tsuboyama_2023_2M0Y": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "F7YBW8_MESOW_Aakre_2015": {"classify_epistasis": {"sample_cut_prob": 0.1, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "F7YBW8_MESOW_Ding_2023": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "GCN4_YEAST_Staller_2018": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "GFP_AEQVI_Sarkisyan_2016": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 1.149052, "diagnostics": {"rows": 17603, "variables": 233, "sparse_mutations": 37912, "candidate_position_pairs": 9131, "complete_square_contexts": 10196}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 1.149052, "diagnostics": {"rows": 17603, "variables": 233, "sparse_mutations": 37912, "candidate_position_pairs": 9131, "complete_square_contexts": 10196}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "GRB2_HUMAN_Faure_2021": {"classify_epistasis": {"requested_cut": 0.75, "sample_cut_prob": 0.75, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "HECD1_HUMAN_Tsuboyama_2023_3DKM": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "HIS7_YEAST_Pokusaeva_2019": {"classify_epistasis": {"seed": 20261004, "time_budget": 15.0, "cutoff_request": "auto", "resolved_cutoff_not_logged": true}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman", "distance": "Hamming to nearest maximizing global optimum", "implementation": "chunked exact comparison of retained uint8 configuration rows"}, "global_optima_accessibility": {}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "degree-bucketed exact GraphFLA nonfocal-moment kernel"}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 305.945553, "diagnostics": {"rows": 493435, "variables": 171, "sparse_mutations": 3429001, "candidate_position_pairs": 1058, "complete_square_contexts": 3828045}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 305.945553, "diagnostics": {"rows": 493435, "variables": 171, "sparse_mutations": 3429001, "candidate_position_pairs": 1058, "complete_square_contexts": 3828045}}}, "ISDH_STAAW_Tsuboyama_2023_2LHR": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "MYO3_YEAST_Tsuboyama_2023_2BTT": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "NKX31_HUMAN_Tsuboyama_2023_2L9R": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "OBSCN_HUMAN_Tsuboyama_2023_1V1C": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "PABP_YEAST_Melamed_2013": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "PHOT_CHLRE_Chen_2023": {"classify_epistasis": {"sample_cut_prob": 0.5, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 164.615181, "diagnostics": {"rows": 167529, "variables": 118, "sparse_mutations": 1336844, "candidate_position_pairs": 105, "complete_square_contexts": 3039183}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 164.615181, "diagnostics": {"rows": 167529, "variables": 118, "sparse_mutations": 1336844, "candidate_position_pairs": 105, "complete_square_contexts": 3039183}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "POLG_PESV_Tsuboyama_2023_2MXD": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "Q6WV12_9MAXI_Somermeyer_2022": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 1.367121, "diagnostics": {"rows": 19302, "variables": 221, "sparse_mutations": 39718, "candidate_position_pairs": 10602, "complete_square_contexts": 11140}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 1.367121, "diagnostics": {"rows": 19302, "variables": 221, "sparse_mutations": 39718, "candidate_position_pairs": 10602, "complete_square_contexts": 11140}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all"}, "global_optima_accessibility": {}}, "Q8WTC7_9CNID_Somermeyer_2022": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 1.688239, "diagnostics": {"rows": 19288, "variables": 237, "sparse_mutations": 44726, "candidate_position_pairs": 10695, "complete_square_contexts": 13524}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 1.688239, "diagnostics": {"rows": 19288, "variables": 237, "sparse_mutations": 44726, "candidate_position_pairs": 10695, "complete_square_contexts": 13524}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "RASK_HUMAN_Weng_2022_abundance": {"classify_epistasis": {"requested_cut": 0.75, "sample_cut_prob": 0.75, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 4.270743, "diagnostics": {"rows": 26011, "variables": 187, "sparse_mutations": 48957, "candidate_position_pairs": 1419, "complete_square_contexts": 1419}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 4.270743, "diagnostics": {"rows": 26011, "variables": 187, "sparse_mutations": 48957, "candidate_position_pairs": 1419, "complete_square_contexts": 1419}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "RASK_HUMAN_Weng_2022_binding-DARPin_K55": {"classify_epistasis": {"requested_cut": 0.75, "sample_cut_prob": 0.75, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 3.98528, "diagnostics": {"rows": 24873, "variables": 187, "sparse_mutations": 46662, "candidate_position_pairs": 1443, "complete_square_contexts": 1443}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 3.98528, "diagnostics": {"rows": 24873, "variables": 187, "sparse_mutations": 46662, "candidate_position_pairs": 1443, "complete_square_contexts": 1443}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "SDA_BACSU_Tsuboyama_2023_1PV0": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "SPA_STAAU_Tsuboyama_2023_1LP1": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "SPG1_STRSG_Olson_2014": {"classify_epistasis": {"requested_cut": 0.75, "sample_cut_prob": 0.75, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 14.720112, "diagnostics": {"rows": 536962, "variables": 55, "sparse_mutations": 1072879, "candidate_position_pairs": 1485, "complete_square_contexts": 1485}}, "gamma_star": {"implementation": "bundled sparse_gamma_statistics", "n_jobs": 1, "computed_jointly": true, "row_subsampling": false, "kernel": "GraphFLA _gamma_grid_moments and _merge_gamma_contributions", "elapsed_sec": 14.720112, "diagnostics": {"rows": 536962, "variables": 55, "sparse_mutations": 1072879, "candidate_position_pairs": 1485, "complete_square_contexts": 1485}}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman", "implementation": "chunked exact Hamming distance to nearest maximizing global optimum", "chunk_rows": 8192, "n_configs": 536962, "n_global_optima": 1, "chunks_processed": 66}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "degree-bucketed exact GraphFLA nonfocal-moment kernel", "improving_edge_pairs": 9256035, "neutral_neighbor_pairs": 1468988, "ordered_pair_denominator": 21450046, "testable_pairs": 10725023, "ee_ordered_pairs": 402099, "timings_seconds": {"pairs_ready": 3.7335254580000026, "moments_ready": 5.567157207999998, "tests_adjusted": 4.580126}, "total_seconds": 13.895389417000004}, "global_optima_accessibility": {}}, "SPG1_STRSG_Wu_2016": {"classify_epistasis": {"sample_cut_prob": 0.75, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "blockwise exact-equivalent GraphFLA EE kernels; full population; symmetric two-orientation BH family", "improving_edge_pairs": 5074250, "neutral_neighbor_pairs": 418479, "ordered_pair_denominator": 10985458, "testable_pairs": 5492729, "edge_block_size": 100000, "equivalence_validation": "public GraphFLA API; two full landscapes; absolute error 0"}, "global_optima_accessibility": {}}, "SPTN1_CHICK_Tsuboyama_2023_1TUD": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "UBE4B_HUMAN_Tsuboyama_2023_3L1X": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "VILI_CHICK_Tsuboyama_2023_1YU5": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "YAIA_ECOLI_Tsuboyama_2023_2KVT": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "YAP1_HUMAN_Araya_2012": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}, "YNZC_BACSU_Tsuboyama_2023_2JVD": {"classify_epistasis": {"sample_cut_prob": null, "seed": 20261004, "time_budget": 15.0}, "diminishing_returns_index": {"method": "pearson"}, "increasing_costs_index": {"method": "pearson"}, "global_idiosyncratic_index": {"n_jobs": 1, "seed": 20261004, "min_pairs": 3}, "r_s_ratio": {}, "gamma": {"n_jobs": 1, "computed_jointly": true}, "gamma_star": {"n_jobs": 1, "computed_jointly": true}, "local_optima_ratio": {}, "autocorrelation": {"walk_length": 20, "walk_times": 1000, "lag": 1, "seed": 20261004}, "fdc": {"method": "spearman"}, "evolvability_enhancing_fraction": {"fdr": 0.01, "effect_type": "all", "implementation": "public GraphFLA API"}, "global_optima_accessibility": {}}}, "resource_audit_by_dms": {"HIS7_YEAST_Pokusaeva_2019": {"retry": "exact full-population graph build and requested feature pass", "max_worker_rss_mib": 4096, "stop_reason": "worker RSS exceeded 4096 MiB (measured 4570 MiB) during fdc; host free memory remained about 64–67%.", "source_release": "v1.3", "n_configs_raw": 496137, "n_graph_configs": 493435, "n_isolated_removed": 2702, "fdc_followup": {"id": "HIS7_YEAST_Pokusaeva_2019", "fdc": -0.2145599946525916, "method": "spearman", "distance": "Hamming to the nearest maximizing global optimum", "n_configs": 493435, "n_global_optima": 1, "n_mutated_positions": 171, "chunk_rows": 8192, "equivalence": "Exact rowwise Hamming and Spearman formula used by GraphFLA FDC; avoids materializing landscape.configs and get_data().", "max_worker_rss_mib": 2137.4, "host_free_percent_at_last_check": 67}, "goa_followup": {"id": "HIS7_YEAST_Pokusaeva_2019", "global_optima_accessibility": 0.04260743562981953, "n_configs": 493435, "n_edges": 3270567, "n_local_optima": 20484, "n_mutated_positions": 171, "method": "public GraphFLA analysis.global_optima_accessibility", "max_worker_rss_mib": 2158.2, "host_free_percent_at_last_check": 66}, "global_idiosyncratic_index_retry": {"max_worker_rss_mib": 3453.0, "host_free_percent_at_last_check": 69, "metric_timeout_sec": 300, "result_details": null}, "evolvability_enhancing_fraction_retry": {"max_worker_rss_mib": 3574.8, "host_free_percent_at_last_check": 66, "metric_timeout_sec": 300, "result_details": {"improving_edge_pairs": 3270567, "neutral_neighbor_pairs": 795336, "ordered_pair_denominator": 8131806, "testable_pairs": 3995250, "ee_ordered_pairs": 9222, "implementation": "degree-bucketed exact GraphFLA nonfocal-moment kernel", "timings_seconds": {"pairs_ready": 1.4526442919999987, "moments_ready": 2.6048977499999992, "tests_adjusted": 1.6239590000000028}, "total_seconds": 5.692149209}}}}, "sampling": {"classify_epistasis": {"seed": 20261004, "time_budget_sec": 15.0, "requested_cut_by_dms": {"A4_HUMAN_Seuma_2022": null, "AMFR_HUMAN_Tsuboyama_2023_4G3O": null, "CAPSD_AAV2S_Sinai_2021": null, "CBX4_HUMAN_Tsuboyama_2023_2K28": null, "CSN4_MOUSE_Tsuboyama_2023_1UFM": null, "D7PM05_CLYGR_Somermeyer_2022": null, "DLG4_HUMAN_Faure_2021": null, "DOCK1_MOUSE_Tsuboyama_2023_2M0Y": null, "F7YBW8_MESOW_Aakre_2015": null, "F7YBW8_MESOW_Ding_2023": null, "GCN4_YEAST_Staller_2018": null, "GFP_AEQVI_Sarkisyan_2016": null, "GRB2_HUMAN_Faure_2021": 0.75, "HECD1_HUMAN_Tsuboyama_2023_3DKM": null, "HIS7_YEAST_Pokusaeva_2019": null, "ISDH_STAAW_Tsuboyama_2023_2LHR": null, "MYO3_YEAST_Tsuboyama_2023_2BTT": null, "NKX31_HUMAN_Tsuboyama_2023_2L9R": null, "OBSCN_HUMAN_Tsuboyama_2023_1V1C": null, "PABP_YEAST_Melamed_2013": null, "PHOT_CHLRE_Chen_2023": null, "POLG_PESV_Tsuboyama_2023_2MXD": null, "Q6WV12_9MAXI_Somermeyer_2022": null, "Q8WTC7_9CNID_Somermeyer_2022": null, "RASK_HUMAN_Weng_2022_abundance": 0.75, "RASK_HUMAN_Weng_2022_binding-DARPin_K55": 0.75, "SDA_BACSU_Tsuboyama_2023_1PV0": null, "SPA_STAAU_Tsuboyama_2023_1LP1": null, "SPG1_STRSG_Olson_2014": 0.75, "SPG1_STRSG_Wu_2016": null, "SPTN1_CHICK_Tsuboyama_2023_1TUD": null, "UBE4B_HUMAN_Tsuboyama_2023_3L1X": null, "VILI_CHICK_Tsuboyama_2023_1YU5": null, "YAIA_ECOLI_Tsuboyama_2023_2KVT": null, "YAP1_HUMAN_Araya_2012": null, "YNZC_BACSU_Tsuboyama_2023_2JVD": null}, "resolved_cut_probability_by_dms": {"A4_HUMAN_Seuma_2022": null, "AMFR_HUMAN_Tsuboyama_2023_4G3O": null, "CAPSD_AAV2S_Sinai_2021": null, "CBX4_HUMAN_Tsuboyama_2023_2K28": null, "CSN4_MOUSE_Tsuboyama_2023_1UFM": null, "D7PM05_CLYGR_Somermeyer_2022": null, "DLG4_HUMAN_Faure_2021": null, "DOCK1_MOUSE_Tsuboyama_2023_2M0Y": null, "F7YBW8_MESOW_Aakre_2015": 0.1, "F7YBW8_MESOW_Ding_2023": null, "GCN4_YEAST_Staller_2018": null, "GFP_AEQVI_Sarkisyan_2016": null, "GRB2_HUMAN_Faure_2021": 0.75, "HECD1_HUMAN_Tsuboyama_2023_3DKM": null, "HIS7_YEAST_Pokusaeva_2019": null, "ISDH_STAAW_Tsuboyama_2023_2LHR": null, "MYO3_YEAST_Tsuboyama_2023_2BTT": null, "NKX31_HUMAN_Tsuboyama_2023_2L9R": null, "OBSCN_HUMAN_Tsuboyama_2023_1V1C": null, "PABP_YEAST_Melamed_2013": null, "PHOT_CHLRE_Chen_2023": 0.5, "POLG_PESV_Tsuboyama_2023_2MXD": null, "Q6WV12_9MAXI_Somermeyer_2022": null, "Q8WTC7_9CNID_Somermeyer_2022": null, "RASK_HUMAN_Weng_2022_abundance": 0.75, "RASK_HUMAN_Weng_2022_binding-DARPin_K55": 0.75, "SDA_BACSU_Tsuboyama_2023_1PV0": null, "SPA_STAAU_Tsuboyama_2023_1LP1": null, "SPG1_STRSG_Olson_2014": 0.75, "SPG1_STRSG_Wu_2016": 0.75, "SPTN1_CHICK_Tsuboyama_2023_1TUD": null, "UBE4B_HUMAN_Tsuboyama_2023_3L1X": null, "VILI_CHICK_Tsuboyama_2023_1YU5": null, "YAIA_ECOLI_Tsuboyama_2023_2KVT": null, "YAP1_HUMAN_Araya_2012": null, "YNZC_BACSU_Tsuboyama_2023_2JVD": null}}, "global_idiosyncratic_index_seed": 20261004, "autocorrelation_seed": 20261004}}, "citation_fields": "publication is the verified peer-reviewed assay paper; proteinGym_reference_doi preserves the benchmark's DOI if it differs"}, "y_format": ".2f", "default_feature": "r_s_ratio", "default_outcome": "esm_1b"}, {"id": "evolution", "features": [{"key": "epistasis.magnitude", "label": "Magnitude epistasis"}, {"key": "epistasis.sign", "label": "Sign epistasis"}, {"key": "epistasis.reciprocal_sign", "label": "Reciprocal sign epistasis"}, {"key": "diminishing_returns_index", "label": "Diminishing returns"}, {"key": "increasing_costs_index", "label": "Increasing costs"}, {"key": "global_idiosyncratic_index", "label": "Global idiosyncrasy"}, {"key": "r_s_ratio", "label": "Roughness-to-slope ratio"}, {"key": "gamma", "label": "Gamma"}, {"key": "gamma_star", "label": "Gamma star"}, {"key": "local_optima_ratio", "label": "Local optima ratio"}, {"key": "autocorrelation", "label": "Autocorrelation"}, {"key": "fdc", "label": "Fitness–distance correlation"}, {"key": "evolvability_enhancing_fraction", "label": "Evolvability-enhancing fraction"}, {"key": "global_optima_accessibility", "label": "Global optimum accessibility"}], "outcomes": [{"key": "rf_greedy", "label": "RF greedy", "metric": "Best fitness percentile"}, {"key": "rf_ucb", "label": "RF uncertainty-guided", "metric": "Best fitness percentile"}, {"key": "random", "label": "Random search", "metric": "Best fitness percentile"}, {"key": "de_greedy", "label": "Greedy directed evolution", "metric": "Best fitness percentile"}], "records": [{"id": "Podgornaia2015_PhoQ", "label": "PhoQ protein-protein interface", "features": {"epistasis.magnitude": 0.4655062619401401, "epistasis.sign": 0.29703530743649614, "epistasis.reciprocal_sign": 0.23745843062336375, "diminishing_returns_index": 0.2132712484822951, "increasing_costs_index": 0.9458304697281467, "global_idiosyncratic_index": 0.7792445436071352, "r_s_ratio": 3.157288262106178, "gamma": 0.22956096920742755, "gamma_star": 0.3160564713841114, "local_optima_ratio": 0.005024303109232335, "autocorrelation": 0.24167036098103692, "fdc": -0.19626581226983875, "evolvability_enhancing_fraction": 0.011434578149747663, "global_optima_accessibility": 0.9751702640961591}, "outcomes": {"rf_greedy": 0.9997779628087704, "rf_ucb": 0.9996769074204546, "random": 0.9979425976928059, "de_greedy": 0.9858550922664163}, "publication": {"title": "Pervasive degeneracy and epistasis in a protein-protein interface", "journal": "Science", "year": 2015, "doi": "10.1126/science.1257360", "url": "https://doi.org/10.1126/science.1257360"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Podgornaia2015_PhoQ.csv"}, {"id": "Wu2016_GB1", "label": "Protein G B1 domain (GB1)", "features": {"epistasis.magnitude": 0.42727289254094936, "epistasis.sign": 0.34977366516988745, "epistasis.reciprocal_sign": 0.2229534422891632, "diminishing_returns_index": 0.2767920280577574, "increasing_costs_index": 0.9228561013224423, "global_idiosyncratic_index": 0.6094700728743864, "r_s_ratio": 2.4771874479570792, "gamma": 0.45087584922619023, "gamma_star": 0.25496822125046453, "local_optima_ratio": 0.0012319146229604783, "autocorrelation": 0.38310695847985377, "fdc": -0.1230743944529954, "evolvability_enhancing_fraction": 0.009554225937385362, "global_optima_accessibility": 0.9975629515067521}, "outcomes": {"rf_greedy": 0.9999832620295794, "rf_ucb": 0.9999819229919458, "random": 0.9990620041376262, "de_greedy": 0.9998734609436198}, "publication": {"title": "Adaptation in protein fitness landscapes is facilitated by indirect paths", "journal": "eLife", "year": 2016, "doi": "10.7554/eLife.16965", "url": "https://doi.org/10.7554/eLife.16965"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Wu2016_GB1.csv"}, {"id": "Jalal2020_NBS", "label": "Noc DNA-binding specificity landscape", "features": {"epistasis.magnitude": 0.37089980161426206, "epistasis.sign": 0.3058129738837405, "epistasis.reciprocal_sign": 0.32328722450199743, "diminishing_returns_index": -0.04092052301678375, "increasing_costs_index": 0.6861410982662656, "global_idiosyncratic_index": 0.6488514024859383, "r_s_ratio": 3.7424148729778834, "gamma": 0.35124369783185994, "gamma_star": 0.1465135662708779, "local_optima_ratio": 0.00360625, "autocorrelation": 0.30730710847593123, "fdc": -0.31301436170743224, "evolvability_enhancing_fraction": 0.019312253289473683, "global_optima_accessibility": 0.9933625}, "outcomes": {"rf_greedy": 0.99999625, "rf_ucb": 0.9999549999999999, "random": 0.99720375, "de_greedy": 0.99806}, "publication": {"title": "Diversification of DNA-Binding Specificity by Permissive and Specificity-Switching Mutations in the ParB/Noc Protein Family", "journal": "Cell Reports", "year": 2020, "doi": "10.1016/j.celrep.2020.107928", "url": "https://doi.org/10.1016/j.celrep.2020.107928"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Jalal2020_NBS.csv"}, {"id": "Jalal2020_parS", "label": "ParB DNA-binding specificity landscape", "features": {"epistasis.magnitude": 0.40235742083601644, "epistasis.sign": 0.31301621076798164, "epistasis.reciprocal_sign": 0.2846263683960019, "diminishing_returns_index": -0.04157259302952305, "increasing_costs_index": 0.6636443911692544, "global_idiosyncratic_index": 0.584570890501797, "r_s_ratio": 3.414783108504666, "gamma": 0.4121143569558325, "gamma_star": 0.17105302858229124, "local_optima_ratio": 0.004425, "autocorrelation": 0.42236875970960625, "fdc": -0.3097150647786003, "evolvability_enhancing_fraction": 0.02041957236842105, "global_optima_accessibility": 0.9928625}, "outcomes": {"rf_greedy": 0.99992375, "rf_ucb": 0.99995625, "random": 0.9981512499999999, "de_greedy": 0.9902993749999999}, "publication": {"title": "Diversification of DNA-Binding Specificity by Permissive and Specificity-Switching Mutations in the ParB/Noc Protein Family", "journal": "Cell Reports", "year": 2020, "doi": "10.1016/j.celrep.2020.107928", "url": "https://doi.org/10.1016/j.celrep.2020.107928"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Jalal2020_parS.csv"}, {"id": "Tu2022_TEV", "label": "TEV protease activity landscape", "features": {"epistasis.magnitude": 0.33442195600597524, "epistasis.sign": 0.3546469054802144, "epistasis.reciprocal_sign": 0.3109311385138104, "diminishing_returns_index": -0.7612229203191835, "increasing_costs_index": 0.4548682139539212, "global_idiosyncratic_index": 0.9209279341147913, "r_s_ratio": 8.835542129191147, "gamma": 0.034292120821619636, "gamma_star": 0.0246356530837724, "local_optima_ratio": 0.0069690571349571425, "autocorrelation": 0.03913981082990978, "fdc": 0.007929723923178013, "evolvability_enhancing_fraction": 0.002115047972113612, "global_optima_accessibility": 0.9436945428952065}, "outcomes": {"rf_greedy": 0.998949299952241, "rf_ucb": 0.998970665862303, "random": 0.9979966317271197, "de_greedy": 0.9901358620516302}, "publication": {"title": "An ultra-high-throughput method for measuring biomolecular activities", "journal": "bioRxiv", "year": 2024, "doi": "10.1101/2022.03.09.483646", "url": "https://doi.org/10.1101/2022.03.09.483646"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Tu2022_TEV.csv"}, {"id": "Tu2022_T7", "label": "T7 RNA polymerase activity (three-site library)", "features": {"epistasis.magnitude": 0.3202586322539103, "epistasis.sign": 0.3375097255385936, "epistasis.reciprocal_sign": 0.342231642207496, "diminishing_returns_index": -0.4639634947138286, "increasing_costs_index": 0.5846222293386543, "global_idiosyncratic_index": 0.9643080388708604, "r_s_ratio": 8.358894103779617, "gamma": 0.018650336197428043, "gamma_star": 0.009708968441894311, "local_optima_ratio": 0.01754646840148699, "autocorrelation": -0.03918735852186762, "fdc": 0.0361740284771143, "evolvability_enhancing_fraction": 0.0003101905111819162, "global_optima_accessibility": 0.8446096654275093}, "outcomes": {"rf_greedy": 0.997888475836431, "rf_ucb": 0.9986765799256505, "random": 0.9982156133828995, "de_greedy": 0.9913605947955391}, "publication": {"title": "An ultra-high-throughput method for measuring biomolecular activities", "journal": "bioRxiv", "year": 2024, "doi": "10.1101/2022.03.09.483646", "url": "https://doi.org/10.1101/2022.03.09.483646"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Tu2022_T7.csv"}, {"id": "Lite2020_ParD2", "label": "ParD3 variants assayed with ParE2", "features": {"epistasis.magnitude": 0.5630717409289682, "epistasis.sign": 0.31345585628176337, "epistasis.reciprocal_sign": 0.12347240278926841, "diminishing_returns_index": -0.059253238649355305, "increasing_costs_index": 0.6003518376033102, "global_idiosyncratic_index": 0.32278049996196384, "r_s_ratio": 0.981872983570605, "gamma": 0.804909568795856, "gamma_star": 0.46786004961423083, "local_optima_ratio": 0.0006343567622430855, "autocorrelation": 0.5604139396903973, "fdc": -0.3015370991821652, "evolvability_enhancing_fraction": 0.02350235407234202, "global_optima_accessibility": 0.9994925145902055}, "outcomes": {"rf_greedy": 1.0, "rf_ucb": 1.0, "random": 0.9976655671149455, "de_greedy": 1.0}, "publication": {"title": "Uncovering the basis of protein-protein interaction specificity with a combinatorially complete library", "journal": "eLife", "year": 2020, "doi": "10.7554/eLife.60924", "url": "https://doi.org/10.7554/eLife.60924"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Lite2020_ParD2.csv"}, {"id": "Lite2020_ParD3", "label": "ParD3 variants assayed with ParE3", "features": {"epistasis.magnitude": 0.6334180356931804, "epistasis.sign": 0.2927372650986881, "epistasis.reciprocal_sign": 0.07384469920813143, "diminishing_returns_index": -0.178857190967055, "increasing_costs_index": 0.4888473728667851, "global_idiosyncratic_index": 0.32012699888604207, "r_s_ratio": 0.8177531063375443, "gamma": 0.7942525095760075, "gamma_star": 0.5846135379318556, "local_optima_ratio": 0.0005074854097944684, "autocorrelation": 0.5454590413134888, "fdc": -0.2451534521151054, "evolvability_enhancing_fraction": 0.02061111086033106, "global_optima_accessibility": 0.9996193859426542}, "outcomes": {"rf_greedy": 1.0, "rf_ucb": 1.0, "random": 0.997526008627252, "de_greedy": 0.9998477543770617}, "publication": {"title": "Uncovering the basis of protein-protein interaction specificity with a combinatorially complete library", "journal": "eLife", "year": 2020, "doi": "10.7554/eLife.60924", "url": "https://doi.org/10.7554/eLife.60924"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Lite2020_ParD3.csv"}, {"id": "Johnston2024_TrpB4", "label": "TrpB active site (four variable residues)", "features": {"epistasis.magnitude": 0.3564145673454999, "epistasis.sign": 0.3514385660438756, "epistasis.reciprocal_sign": 0.2921468666106245, "diminishing_returns_index": 0.1596290459248483, "increasing_costs_index": 0.8708697852013935, "global_idiosyncratic_index": 0.6552707990139779, "r_s_ratio": 4.362459443490163, "gamma": 0.394165626024921, "gamma_star": 0.07375041376715637, "local_optima_ratio": 0.005039936152429789, "autocorrelation": 0.36559064656808565, "fdc": -0.11155100086027349, "evolvability_enhancing_fraction": 0.014171187479358617, "global_optima_accessibility": 0.9875635490702512}, "outcomes": {"rf_greedy": 0.9998849989631055, "rf_ucb": 0.999896310540505, "random": 0.9965273457383633, "de_greedy": 0.9945987217917539}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB4.csv"}, {"id": "Johnston2024_TrpB3A", "label": "TrpB active site (three-site library A)", "features": {"epistasis.magnitude": 0.38375495048082175, "epistasis.sign": 0.3602513033631656, "epistasis.reciprocal_sign": 0.2559937461560127, "diminishing_returns_index": -0.4227443370317522, "increasing_costs_index": 0.6261390394743028, "global_idiosyncratic_index": 0.866197438477808, "r_s_ratio": 2.17631627394597, "gamma": 0.11828562757186226, "gamma_star": 0.12728387004252503, "local_optima_ratio": 0.007778195960356292, "autocorrelation": 0.1393994734254282, "fdc": -0.17765764365707418, "evolvability_enhancing_fraction": 0.020674208924366853, "global_optima_accessibility": 0.9627399322544222}, "outcomes": {"rf_greedy": 0.9996487266340484, "rf_ucb": 0.9997365449755364, "random": 0.9985196336720612, "de_greedy": 0.9978296324175135}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3A.csv"}, {"id": "Johnston2024_TrpB3B", "label": "TrpB active site (three-site library B)", "features": {"epistasis.magnitude": 0.35187560297372455, "epistasis.sign": 0.3553317064865785, "epistasis.reciprocal_sign": 0.292792690539697, "diminishing_returns_index": -0.4648359709648337, "increasing_costs_index": 0.6439560862372854, "global_idiosyncratic_index": 0.9394482177647541, "r_s_ratio": 3.7341779365436087, "gamma": 0.051622250494064825, "gamma_star": 0.056821220237682206, "local_optima_ratio": 0.009254627313656828, "autocorrelation": 0.06783760766853235, "fdc": -0.02113756301484364, "evolvability_enhancing_fraction": 0.005156449623089655, "global_optima_accessibility": 0.9501000500250125}, "outcomes": {"rf_greedy": 0.9991745872936468, "rf_ucb": 0.9996498249124564, "random": 0.9975862931465734, "de_greedy": 0.9912581290645323}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3B.csv"}, {"id": "Johnston2024_TrpB3C", "label": "TrpB active site (three-site library C)", "features": {"epistasis.magnitude": 0.36419643724603806, "epistasis.sign": 0.35745840881944485, "epistasis.reciprocal_sign": 0.2783451539345171, "diminishing_returns_index": -0.5177447569171416, "increasing_costs_index": 0.5263074275930909, "global_idiosyncratic_index": 0.8989979016182247, "r_s_ratio": 3.4378009784655226, "gamma": 0.07758805877630812, "gamma_star": 0.07513314844336687, "local_optima_ratio": 0.007505629221916437, "autocorrelation": 0.11014321389919898, "fdc": -0.06491360727118037, "evolvability_enhancing_fraction": 0.011954229992093474, "global_optima_accessibility": 0.9310733049787341}, "outcomes": {"rf_greedy": 0.9992494370778084, "rf_ucb": 0.999074305729297, "random": 0.9967850888166124, "de_greedy": 0.9958218663998}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3C.csv"}, {"id": "Johnston2024_TrpB3D", "label": "TrpB active site (three-site library D)", "features": {"epistasis.magnitude": 0.38257989393729147, "epistasis.sign": 0.3583145803041912, "epistasis.reciprocal_sign": 0.2591055257585173, "diminishing_returns_index": -0.03174445032996317, "increasing_costs_index": 0.8462930119108031, "global_idiosyncratic_index": 0.7765870822340654, "r_s_ratio": 2.212572518717743, "gamma": 0.2609109958906258, "gamma_star": 0.11602060949875444, "local_optima_ratio": 0.005925544248357594, "autocorrelation": 0.2648172966750049, "fdc": -0.1250026427244742, "evolvability_enhancing_fraction": 0.011763535123976505, "global_optima_accessibility": 0.9853149555584181}, "outcomes": {"rf_greedy": 0.9999871183820688, "rf_ucb": 0.9999613551462063, "random": 0.9971402808192709, "de_greedy": 0.9769676671389925}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3D.csv"}, {"id": "Johnston2024_TrpB3E", "label": "TrpB active site (three-site library E)", "features": {"epistasis.magnitude": 0.3385266684258229, "epistasis.sign": 0.34670674483439073, "epistasis.reciprocal_sign": 0.31476658673978636, "diminishing_returns_index": -0.18098721408848736, "increasing_costs_index": 0.8641985023982354, "global_idiosyncratic_index": 0.925835610945393, "r_s_ratio": 8.066350688202487, "gamma": 0.13307595355817567, "gamma_star": 0.02073754753350784, "local_optima_ratio": 0.011551983927674536, "autocorrelation": 0.13132581065920232, "fdc": -0.036235217276366985, "evolvability_enhancing_fraction": 0.0006611796860833838, "global_optima_accessibility": 0.9519085886489201}, "outcomes": {"rf_greedy": 0.9978653942742339, "rf_ucb": 0.997388247112004, "random": 0.9975640381717732, "de_greedy": 0.9900929181315922}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3E.csv"}, {"id": "Johnston2024_TrpB3F", "label": "TrpB active site (three-site library F)", "features": {"epistasis.magnitude": 0.3386559306286776, "epistasis.sign": 0.3517126045215237, "epistasis.reciprocal_sign": 0.3096314648497987, "diminishing_returns_index": -0.5211139976479823, "increasing_costs_index": 0.5868789284133726, "global_idiosyncratic_index": 0.9209911507954616, "r_s_ratio": 8.928644040682112, "gamma": 0.08761917912460393, "gamma_star": 0.03081749648525203, "local_optima_ratio": 0.011115419413209255, "autocorrelation": 0.10815797546039714, "fdc": -0.19740880683474402, "evolvability_enhancing_fraction": 0.0038031507820957926, "global_optima_accessibility": 0.9705312136487011}, "outcomes": {"rf_greedy": 1.0, "rf_ucb": 1.0, "random": 0.9989013829649735, "de_greedy": 0.9906165180302443}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3F.csv"}, {"id": "Johnston2024_TrpB3G", "label": "TrpB active site (three-site library G)", "features": {"epistasis.magnitude": 0.3394560882732152, "epistasis.sign": 0.3507660939701958, "epistasis.reciprocal_sign": 0.309777817756589, "diminishing_returns_index": -0.23785182201265304, "increasing_costs_index": 0.820524976341838, "global_idiosyncratic_index": 0.850660856860407, "r_s_ratio": 4.309679563722612, "gamma": 0.15522755419350587, "gamma_star": 0.03636465939114782, "local_optima_ratio": 0.009631225446711444, "autocorrelation": 0.1296196137500479, "fdc": -0.17556773279160104, "evolvability_enhancing_fraction": 0.001971880444618522, "global_optima_accessibility": 0.9735141300215435}, "outcomes": {"rf_greedy": 1.0, "rf_ucb": 1.0, "random": 0.9980103915853504, "de_greedy": 0.986845773666202}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3G.csv"}, {"id": "Johnston2024_TrpB3H", "label": "TrpB active site (three-site library H)", "features": {"epistasis.magnitude": 0.33116720819795054, "epistasis.sign": 0.34690077480629844, "epistasis.reciprocal_sign": 0.3219320169957511, "diminishing_returns_index": -0.22143468117543183, "increasing_costs_index": 0.8785559338813973, "global_idiosyncratic_index": 1.027150749195134, "r_s_ratio": 9.712154481040322, "gamma": 0.04779949730281115, "gamma_star": 0.009375599487219567, "local_optima_ratio": 0.014515292897874546, "autocorrelation": 0.011092355139876296, "fdc": -0.011938138573382337, "evolvability_enhancing_fraction": 0.00018145949265786977, "global_optima_accessibility": 0.944530844997408}, "outcomes": {"rf_greedy": 0.998535510627268, "rf_ucb": 0.999559357179886, "random": 0.9983022291342664, "de_greedy": 0.9841368584758943}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3H.csv"}, {"id": "Johnston2024_TrpB3I", "label": "TrpB active site (three-site library I)", "features": {"epistasis.magnitude": 0.4795443636229369, "epistasis.sign": 0.35577053256040075, "epistasis.reciprocal_sign": 0.16468510381666235, "diminishing_returns_index": 0.19511769735473922, "increasing_costs_index": 0.8346365679478104, "global_idiosyncratic_index": 0.577680287710991, "r_s_ratio": 1.5083242326598092, "gamma": 0.5001167844790301, "gamma_star": 0.3002139697959707, "local_optima_ratio": 0.0020554984583761563, "autocorrelation": 0.4032644076571373, "fdc": -0.24031617096271166, "evolvability_enhancing_fraction": 0.042577760533584606, "global_optima_accessibility": 0.9973021582733813}, "outcomes": {"rf_greedy": 1.0, "rf_ucb": 1.0, "random": 0.9986510791366907, "de_greedy": 0.9923432682425488}, "publication": {"title": "A combinatorially complete epistatic fitness landscape in an enzyme active site", "journal": "Proceedings of the National Academy of Sciences", "year": 2024, "doi": "10.1073/pnas.2400439121", "url": "https://doi.org/10.1073/pnas.2400439121"}, "source_url": "https://github.com/COLA-Laboratory/GraphFLA/blob/main/data/BioSequence/Johnston2024_TrpB3I.csv"}], "provenance": {"kind": "GraphFLA directed-evolution benchmark on empirical protein amino-acid landscapes", "landscape_count": 18, "feature_seed": 20261004, "simulation_seeds": [20261004, 20261005, 20261006, 20261007, 20261008, 20261009, 20261010, 20261011, 20261012, 20261013], "simulation_repeats": 10, "outcome_aggregation": "Arithmetic mean of the 10 seed-level best-fitness percentiles for each landscape and strategy.", "query_budget_cap": 480, "exact_budget_strategies": ["rf_greedy", "rf_ucb", "random"], "early_stopping_strategy": "de_greedy", "initial_random_measurements": 96, "active_learning_rounds": 4, "active_learning_batch_size": 96, "strategy_protocols": {"rf_greedy": "96 shared random observations, then four batches of 96 unmeasured genotypes ranked by a 32-tree RandomForestRegressor predictive mean.", "rf_ucb": "Same 96+4x96 schedule and forest; rank by standardized predictive mean plus one between-tree standard deviation.", "random": "480 unique uniformly random observed genotypes; shares the initial 96 with both RF strategies.", "de_greedy": "Starts at the first shared initial genotype, assays all available unmeasured one-site neighbors, moves to the best improving neighbor, and stops at a local optimum or the 480-query cap. Actual counts are in simulation_runs.csv."}, "random_forest": {"n_estimators": 32, "max_features": "sqrt", "max_depth": 10, "min_samples_leaf": 1, "bootstrap": true, "n_jobs": 1, "target_standardization": "mean and population SD of currently observed labels at each round", "uncertainty": "population SD of individual-tree predictions"}, "fitness_rank": "fraction of observed genotypes with fitness <= best fitness measured; ties share the same empirical percentile", "objective": "maximize the source-file fitness column without rescaling; ranks are computed against all observed genotypes in that landscape", "source_population": "Every row in each listed BioSequence CSV is retained; unmeasured sequence combinations are not imputed or removed.", "features_source": "GraphFLA landscape analysis on the full retained one-edit graph with epsilon=0; parameter settings are in README.md."}, "y_format": "%", "default_feature": "epistasis.reciprocal_sign", "default_outcome": "rf_greedy"}]