Noves aproximacions per a la identificació in silico de variants patogèniques en els gens de predisposició al càncer hereditari de mama i ovari brca1 i brca2

Abstract

Germline variants in BRCA1 and BRCA2 can disrupt the DNA protective role of these proteins resulting in an increased risk of developing hereditary breast and ovarian cancer (HBOC). Identification of those individuals carrying pathogenic variants will allow channeling them into specific programs of prevention and surveillance, incrementing their survival rates. For this purpose, first, it is necessary to identify which of the variants are pathogenic. Unfortunately, there is not always enough information to reach a conclusion. In this situation, pathogenicity predictors designed to computationally estimate the damage caused by variants, can provide valuable information. Here, we present a novel family of pathogenicity predictors for BRCA1 and BRCA2. These predictors differ in their objective: one is trained to estimate the molecular impact of variants on the HDR function of BRCA1 and BRCA2, and the other is trained to estimate the clinical significance of a variant, that is, whether it should be classified as pathogenic or neutral. Their performances have been tested and are comparable to those of widely used predictors in the field. Additionally, we presented them to the ENIGMA challenge from the 5th Critical Assessment of Genome Interpretation (CAGI), finding that our predictors, especially those estimating the functional impact of variants, ranked in the top positions compared to other tools. In order to disseminate this family of predictors to the scientific community, we have built the BRASS website (https://www.biotoclin.org/BRASS), where users can analyze their missense BRCA1 and BRCA2 variants. More advanced users can also interpret the predictions using a reliability metric and several plots contextualizing the score to that of a set of manually curated variants. Independently, we applied our knowledge about pathogenicity predictors in a large international effort to characterize a novel pediatric neurologic disorder caused by pathogenic variants in histone H3.3. We combined the use of standard pathogenic predictors with evidence from structural analyses and biophysical computations to provide a mechanistic view of the impact of the causative variants.