Bayesian Inference for Mixed Model-Based Genome-Wide Analysis of Expression Quantitative Trait Loci by Gibbs Sampling

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Expression quantitative trait loci (eQTLs) are crucial for understanding the genetic basis of cellular activities and complex phenotypes.
• Mixed models are effective for eQTL identification by accounting for polygenic effects, treating them as random variables.
• Both frequentist and Bayesian approaches exist for mixed model-based eQTL analysis.

Purpose of the Study:

• To review mixed model-based Bayesian eQTL analysis using Gibbs sampling.
• To discuss theoretical and practical aspects of Bayesian inference in this context.
• To highlight the strengths and utility of Bayesian inference for eQTL studies.

Main Methods:

• Utilizes mixed models to explain polygenic effects in eQTL analysis.
• Employs Bayesian inference with Gibbs sampling (a Markov chain Monte Carlo method) to estimate parameters.
• Focuses on marginal posterior distributions for parameter estimation.

Main Results:

• Bayesian inference provides posterior probability distributions reflecting parameter uncertainty, beneficial for small sample sizes.
• Gibbs sampling offers a solution for the challenge of marginalizing joint posterior distributions.
• The Bayesian approach incorporates prior knowledge, making it valuable with accumulating eQTL data.

Conclusions:

• Mixed model-based Bayesian eQTL analysis using Gibbs sampling is a robust statistical framework.
• This approach is particularly advantageous when frequentist methods are limited by small sample sizes.
• Widespread adoption of this Bayesian method is expected to accelerate the understanding of eQTLs and their regulatory functions.