Screen-VarCal: An Interpretable Probabilistic Framework for Recalibrating ACMG Rule-Based Variant Classification in Preventive Medicine

Abstract

Motivationhole-genome sequencing (WGS) is increasingly used for preventive genomics, yet rule-based ACMG engines such as InterVar were tuned for high pre-test probability diagnostics. In screening contexts, these heuristics can inflate pathogenic/likely pathogenic (P/LP) calls, prompting unnecessary follow-up. We sought an interpretable, data-driven recalibration tailored to proactive use. ResultsAcross 20 WGS cases, InterVar flagged 109 variants as P/LP; only 18 (16.5%) were concordant with ClinVar P/LP assertions. The remaining 83.5% were largely absent from Clin-Var (n=68) or mapped to benign/likely benign (n=13). Nearly all flagged variants (96%) were heterozygous, predominantly in autosomal recessive genes (e.g., FAM20C, MTMR2 ), indicating a dominant carrier inflation effect; recurrent loci (e.g., ATXN3, FAM20C ) further amplified yield. We introduce Screen-VarCal, an interpretable probabilistic framework that combines logistic regression with isotonic adjustment to align InterVar outputs with observed ClinVar P/LP distributions. Screen-VarCal reduced false positives by[~] 60% while retaining all ClinVar-concordant findings and yields calibrated probabilities with coefficients that transparently link ACMG evidence categories (PVS/PS/PM/PP/BS/BP) and zygosity to risk.