Genetic variant selection: learning across traits and sites

TL;DR

This paper introduces Bayesian methods with novel priors for selecting genetic variants across traits and sites, improving prioritization in resequencing studies by leveraging correlations and shared information.

Contribution

It proposes two new prior distributions within a Bayesian multivariate linear regression framework to enhance variant prioritization by borrowing evidence across phenotypes and mutations.

Findings

Simulations show improved variant detection accuracy.

Re-analysis of sequencing data demonstrates practical benefits.

Bayesian approach effectively integrates multiple sources of evidence.

Abstract

We consider resequencing studies of associated loci and the problem of prioritizing sequence variants for functional follow-up. Working within the multivariate linear regression framework helps us to account for correlation across variants, and adopting a Bayesian approach naturally leads to posterior probabilities that incorporate all information about the variants' function. We describe two novel prior distributions that facilitate learning the role of each variant by borrowing evidence across phenotypes and across mutations in the same gene. We illustrate their potential advantages with simulations and re-analyzing a dataset of sequencing variants.