MENDELSEEK Predicts Mendelian Genes And Mechanisms
Zhou, Edelman, and Skolnick publish MENDELSEEK on February 17, 2026, in PLoS Computational Biology, introducing a machine-learning framework that predicts Mendelian genes by integrating residue variation scores, pathway participation, GO processes, and protein language model features. In 10-fold cross-validation across 16,946 genes MENDELSEEK achieved AUC 0.869 and AUPR 0.737, predicted 1,277 novel candidates with precision >0.7, and found Mendelian genes are interaction-rich and evolutionarily ancient.
Key Points
- 1Achieved high predictive performance (AUC 0.869, AUPR 0.737), outperforming ENTPRISE+ENTPRISE-X and REVEL on 16,946 genes
- 2Integrates residue variation scores, pathway membership, Gene Ontology processes, and protein-language-model features to capture Mendelian gene biology
- 3Predicts 1,277 novel Mendelian gene candidates with precision >0.7, guiding geneticists in variant and gene prioritization
Scoring Rationale
Strong methodological advance and validated genome-wide predictions, supported by peer-reviewed publication and open code, but requires clinical validation.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
