A Bayesian test to identify variance effects

TL;DR

This paper introduces a Bayesian heteroskedastic linear regression model, BTH, to detect genetic variants affecting trait variance, outperforming classical tests and applied to methylation and expression QTL data.

Contribution

The paper presents a novel Bayesian test for heteroskedasticity that improves detection of variance effects in quantitative traits over existing methods.

Findings

BTH outperforms classical heteroskedasticity tests in simulations.

Application to QTL data identifies variance QTLs effectively.

The approach reduces overfitting by incorporating uncertainty.

Abstract

Identifying genetic variants that regulate quantitative traits, or QTLs, is the primary focus of the field of statistical genetics. Most current methods are limited to identifying mean effects, or associations between genotype and the mean value of a quantitative trait. It is possible, however, that a genetic variant may affect the variance of the quantitative trait in lieu of, or in addition to, affecting the trait mean. Here, we develop a general methodological approach to identifying covariates with variance effects on a quantitative trait using a Bayesian heteroskedastic linear regression model. We show that our Bayesian test for heteroskedasticity (BTH) outperforms classical tests for differences in variation across a large range of simulations drawn from scenarios common to the analysis of quantitative traits. We apply BTH to methylation QTL study data and expression QTL study data to identify variance QTLs. When compared with three tests for heteroskedasticity used in genomics, we illustrate the benefits of using our approach, including avoiding overfitting by incorporating uncertainty and flexibly identifying heteroskedastic effects.