From digenic combinations to oligogenic networks via a new predictive approach

Research output: Contribution to journalMeeting abstract (Journal)

Abstract

Abstract (max. 300 words)
The wide-spread use of high throughput-sequencing in the last decades has helped immensely in unravelling the genetic architecture of many rare diseases as wells as identifying the causative variant in Mendelian ones. Notwithstanding the many successes, different issues remain unsolved: Certain cases exhibit incomplete penetrance, phenotypic variability, locus heterogeneity, as well as non-Mendelian patterns of inheritance. Novel technological advancements are thus required to move beyond this current state of the art. Our team pushes the envelope on new pathogenicity prediction and ranking tools for the identification of the oligogenic causes of diseases. We aim to identify more complex inheritance models, where variants in a small number of genes cause or modulate the development of disease.

In this presentation I discuss two advancements we recently published: a new pathogenicity predictor, developed using published medical data on digenic diseases, and an online, publicly available platform that takes the results of this predictor, aggregates this into oligogenic networks, which combine highly affected gene pairs, and contextualises these results with molecular knowledge. This platform, called the Oligogenic resource for variant analysis (or ORVAL), allows researchers to perform predictions on gene panels as well as full exomes, receiving a full report on the potentially relevant combinations, integrated into gene-pairs and networks, and annotated with known protein relationships as well as functional information for all candidate pathogenic combinations that can be found.

These two advancements provide researchers with a novel way to explore patient exome data while also adding molecular knowledge and other pieces of information. The predictor, which is called the variant combination pathogenicity predictor (or VarCoPP) assigns pathogenicity scores to variant combinations in gene pairs, similarly to well-known monogenic pathogenicity predictors currently used in prioritisation pipelines. Yet, as VarCoPP is trained on digenic diseases data, it identifies other culprits than those that would be identified in a monogenic way, opening new routes for exploration when the classical approaches fail. In this way, we hope to assist in providing more accurate insights into the reasons for phenotypic heterogeneity in diagnosed patients as well as to raise the diagnosis level for many rare diseases.
Original languageEnglish
Pages (from-to)12-12
JournalEuropean Journal of Human Genetics
Volume28
Issue numberSUPPL 1
Publication statusPublished - Dec 2020

Cite this