Integration of VarSome API in an existing bioinformatic pipeline for automated ACMG interpretation of clinical variants.

Abstract

While the bioinformatic workflow, from quality control to annotation, is quite standardized, the interpretation of variants is still a challenge. The decreasing cost of massively parallel NGS has produced hundreds of variants per patient to analyze and interpret. The ACMG "Standards and guidelines for the interpretation of sequence variants", widely adopted in clinical settings, assume that the clinician has a comprehensive knowledge of the literature and the disease. To semi-automatize the application of the guidelines, we decided to develop an algorithm that exploits VarSome, a widely used platform that interprets variants on the basis of information from more than 70 genome databases. Here we explain how we integrated VarSome API into our existing clinical diagnostic pipeline for NGS data to obtain validated reproducible results as indicated by accuracy, sensitivity and specificity. We validated the automated pipeline to be sure that it was doing what we expected. We obtained 100% sensitivity, specificity and accuracy, confirming that it was suitable for use in a diagnostic setting.