MSH6 is a DNA mismatch-repair gene whose loss-of-function variants raise the lifetime risk of colorectal andendometrial cancer. Of the thousands of MSH6 variants in ClinVar, many remain classified as variants of uncertainsignificance (VUS). This project collected all 4,007 uncertain missense MSH6 variants from ClinVar and scored eachone with four publicly available prediction tools — SIFT, CADD, REVEL, and AlphaMissense — combined into a singleconsensus score. After adding population rarity evidence from gnomAD, 388 variants met stringent high-prioritycriteria. Validation on 143 MSH6 variants with established ClinVar classifications (67 pathogenic, 76 benign) gave anAUC of 0.975 (95% bootstrap CI, 0.947–0.995). At a 0.5 cutoff, sensitivity was 0.985, specificity 0.737, precision 0.767,and F1 0.863. AlphaMissense alone achieved a slightly higher AUC (0.980) and better accuracy at the same cutoff;the consensus score was more sensitive but less specific. The combined score is therefore best understood as atransparent prioritization method, not a demonstrated improvement over individual predictors. Real-worldperformance on genuinely novel variants is likely lower, given partial training overlap. The pipeline is implementedas a public web tool, Genevux, and is intended for research prioritization only.
No takes yet. Share an insight, caveat, or question.
Yaseen shah Shah (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: