Mikhail Gudkov

SVlog provides a logic programming framework for analyzing structural variants in genomic disease by applying deterministic rules to genomic data and annotations.

This study benchmarks variant pathogenicity predictors using population genetics data, finding CADD and REVEL as the best performers and highlighting biases in ClinVar-based training.

The paper introduces CAPS, a novel metric for estimating negative selection in the human genome that improves upon MAPS by removing the mutability layer, resulting in less biased estimates and more accurate confidence intervals.