Simultaneous inferences based on empirical Bayes methods and false discovery rates ineQTL data analysis

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) identify genetic variants linked to complex diseases.
• Expression quantitative trait loci (eQTL) mapping reveals functional effects of single nucleotide polymorphisms (SNPs).
• eQTL studies involve massive multiple testing (~10^6), posing computational and statistical challenges.

Purpose of the Study:

• To address computational and statistical issues in eQTL studies.
• To propose and evaluate an empirical Bayes method for eQTL analysis.
• To compare the proposed method with existing approaches.

Main Methods:

• Utilized a parametric empirical Bayes (PEB) method with a three-component t-mixture model.
• Employed the Expectation/Conditional Maximization Either (ECME) algorithm for inference.
• Compared PEB with a nonparametric empirical Bayes (NPEB) alternative via simulation.

Main Results:

• The PEB method demonstrated an advantage over NPEB.
• Applied to human liver cohort (LHC) data, the method discovered more significant SNPs (FDR<10%) than a previous study.
• The empirical Bayes approach effectively controls the false positive rate.

Conclusions:

• The empirical Bayes method, using local false discovery rate (lfdr), offers an improvement over p-value-based methods for eQTL analysis.
• This approach enhances the discovery of significant SNPs while controlling false positives.
• The proposed methodology provides a robust framework for analyzing large-scale eQTL data.