Pathogenicity of mutation analyzer (PathoMAN): A fast automation of germline genomic variant curation in clinical sequencing.

Abstract

1529 Background: A challenge in clinical oncology is interpretation of multiplexed gene sequencing of patients at risk. The plethora of variants to be curated for pathogenicity or actionability poses a growing burden for cancer care professionals. Current guidelines by the ACMG requires the aggregation of multiple lines of genomic data evidences from diverse resources. A computational tool that automates, provide uniformity and significantly speed the interpretive process is thus necessary. Methods: The Pathogenicity of Mutation Analyzer (PathoMAN), is a tool that automates germline genomic variant curation from clinical sequencing based on ACMG guidelines. PathoMAN aggregates multiple tracks of genomic, protein and disease specific information from public sources such as ClinVar, ExAC, UniProt, 1000 genomes, dbNSFP and locus specific databases. Variant specific and gene specific annotations are used to classify variants to model the ACMG rubric. We analyzed 2500 manually curated and classified, high quality variants in 180 genes from 3 large, published studies to quantify the performance of PathoMAN; analyzing 242 pathogenic/likely pathogenic (P/LP), 1272 benign/likely benign (B/LB) and 1261 variants of uncertain significance (VUS). We report the summary of PathoMAN classifications in four categories contrasted against the manual curation. Results: PathoMan achieves an average of 75% concordance and 1.5% discordance for P/LP mutations and 60% and 0.1% for B/LB variants. PathoMAN is able to resolve 12% of reported VUS as either P/LP or B/LB. It loses resolution to classify 25% of P/LP and B/LB variants due to lack of information and due to inconsistencies in available data from public resources. Conclusions: PathoMAN provides a breakthrough in rapid classification of genetic variants by generation of robust models using a knowledgebase of diverse genetic data. It is easily accessible, web-based resource that allows the community to rapidly test a large number of variants for pathogenicity. Such bioinformatic tools are essential to reduce manual workload of a domain level experts. We propose, a new nosology for the 5 ACMG classes to facilitate better reporting to ClinVar.