Robust identification of regulatory variants (eQTLs) using a differential expression framework developed for...

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Expression quantitative trait loci (eQTL) studies link genetic variants to gene expression, but RNA-sequencing data present normalization and distribution challenges.
• Current methods, like the Genotype-Tissue Expression (GTEx) project, normalize data using trimmed means of M-values (TMM) and inverse normal transformation.
• These transformations may obscure biologically relevant genetic variants.

Purpose of the Study:

• To investigate an alternative eQTL analysis framework using differential gene expression (DGE) models.
• To determine if DGE frameworks can identify eQTLs without prior data transformation to a normal distribution.
• To compare the performance of a DGE framework against the established GTEx framework.

Main Methods:

• Utilized a negative binomial model, suitable for count data, within a DGE framework for eQTL analysis.
• Applied the GTEx framework with ANOVA and additive models for comparison.
• Identified significant eQTLs using a stringent statistical threshold (P < 5 × 10⁻⁸).

Main Results:

• The DGE framework identified substantially more significant eQTLs (930 and 6) compared to the GTEx framework (35 and 39).
• No overlap was observed between the significant eQTLs identified by the DGE and GTEx frameworks.
• The DGE framework successfully identified trans eQTLs, including those exhibiting complete dominance, which were missed by the GTEx approach.

Conclusions:

• RNA-sequencing data transformation to a normal distribution is unnecessary when employing a DGE framework for eQTL analysis.
• The proposed DGE approach enhances the detection of biologically relevant genetic variants.
• This method offers a more sensitive alternative for identifying eQTLs, particularly trans eQTLs.