Protocol metadata evidence: proteingym-paper.xml

What this source contributed

Records in this release that cite this source. A record can cite several sources.

• 2 methods
• 1 benchmark
• 3 protocols
• 9 evidence claims

Claims

Statements this source supports, each with where in the source it was checked.

• EVmutation: description: Description Unsupervised statistical model, fitted to a multiple sequence alignment of the protein family, that predicts mutation effects while explicitly capturing dependencies between residue positions. (qa-descriptions-20261011-a-source-hopf2017: abstractText ('EVmutation, an unsupervised statistical method for predicting the effects of mutations that explicitly captures residue dependencies between positions'); profile-protocol-proteingym-paper-xml-a58110e0: Baselines (alignment-based models 'are trained on Multiple Sequence Alignments'))
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-supervised-substitutions-auc-144fe22 to use-case-brca1-brca2-germline-interpretation: Assessed by ProteinGym clinical substitutions, supervised predictors, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-supervised-substitutions-auc-summary-144fe22: Summary_performance_clinical_substitutions_AUC.csv (supervised))
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-supervised-substitutions-auc-144fe22 to use-case-missense-pp3-bp4-calibrated-evidence: Assessed by ProteinGym clinical substitutions, supervised predictors, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-supervised-substitutions-auc-summary-144fe22: Summary_performance_clinical_substitutions_AUC.csv (supervised); README line 106; profile-protocol-proteingym-paper-xml-a58110e0: ProteinGym NeurIPS 2023 paper (PMC10723403): 'Clinical datasets', Appendix A.3.2, Appendix 'Limitations' ('Bias', 'Circularity'))
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-supervised-substitutions-auc-144fe22 to use-case-rare-disease-candidate-ranking: Assessed by ProteinGym clinical substitutions, supervised predictors, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-supervised-substitutions-auc-summary-144fe22: Summary_performance_clinical_substitutions_AUC.csv (supervised); README line 106)
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-zero-shot-indels-auc-144fe22 to use-case-rare-disease-candidate-ranking: Assessed by ProteinGym clinical indels, zero-shot, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-zero-shot-indels-auc-summary-144fe22: Summary_performance_clinical_indels_AUC.csv; README line 22)
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-zero-shot-substitutions-auc-144fe22 to use-case-brca1-brca2-germline-interpretation: Assessed by ProteinGym clinical substitutions, zero-shot, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-zero-shot-substitutions-auc-summary-144fe22: Summary_performance_clinical_substitutions_AUC.csv (zero-shot))
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-zero-shot-substitutions-auc-144fe22 to use-case-missense-pp3-bp4-calibrated-evidence: Assessed by ProteinGym clinical substitutions, zero-shot, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-zero-shot-substitutions-auc-summary-144fe22: Summary_performance_clinical_substitutions_AUC.csv (zero-shot); README lines 22-28, 106; profile-protocol-proteingym-paper-xml-a58110e0: ProteinGym NeurIPS 2023 paper (PMC10723403): 'Clinical datasets', Appendix A.3.2, Appendix 'Limitations' ('Bias', 'Circularity'))
• Relevance of vep-lit-20261014-protocol-proteingym-clinical-zero-shot-substitutions-auc-144fe22 to use-case-rare-disease-candidate-ranking: Assessed by ProteinGym clinical substitutions, zero-shot, average AUC (commit 144fe22) (vep-lit-20261014-source-proteingym-clinical-zero-shot-substitutions-auc-summary-144fe22: Summary_performance_clinical_substitutions_AUC.csv (zero-shot); README lines 22-28, 106)
• Site-Independent: description: Description Alignment-based model, fitted to a multiple sequence alignment of the protein family, that predicts mutation effects from each position separately, without the dependencies between positions that EVmutation adds. (profile-protocol-proteingym-paper-xml-a58110e0: Baselines ('Alignment-based models, such as site-independent and EVmutation models [Hopf et al., 2017] ... are trained on Multiple Sequence Alignments'); qa-descriptions-20261011-a-source-hopf2017: abstractText (EVmutation 'explicitly captures residue dependencies between positions'; most methods 'do not consider the interdependencies of residues'))

Models, benchmarks and other records

Evidence

Source checking verifies the cited claim or transcription. It does not establish independent reproduction.