Detection of cis-acting regulatory SNPs using allelic expression data

Area of Science:

• Genetics and Genomics
• Statistical Bioinformatics
• Human Molecular Genetics

Background:

• Allelic expression (AE) imbalance between gene alleles can identify cis-acting regulatory SNPs (rSNPs).
• Detecting rSNPs is challenging with phase-unknown data and varying linkage disequilibrium (LD) between rSNPs and transcribed SNPs (tSNPs).

Purpose of the Study:

• To develop and evaluate novel statistical tests for AE analysis in phase-unknown individuals.
• To assess test performance across different levels of LD between rSNPs and tSNPs.

Main Methods:

• Proposed three new AE analysis tests: minimum P-value (F and t tests), combined F and t tests, and a mixture-model-based test.
• Compared proposed tests against existing F, t, and regression-based tests using simulated phase-unknown and phase-known data.
• Evaluated test power under various LD scenarios and the impact of ungenotyped SNPs.

Main Results:

• Test performance ranking strongly depends on the magnitude of LD between rSNPs and tSNPs.
• Proposed tests demonstrated higher power than existing methods for phase-unknown data with moderate LD (∼0.2 to ∼0.8).
• The presence of a second ungenotyped rSNP minimally affected the proposed tests' validity or power rankings.

Conclusions:

• Recommended specific tests based on LD levels for detecting cis-acting rSNPs using phase-unknown AE data: F test for low LD (<0.2), t test for strong LD (<0.7), and mixture-model test for intermediate LD (0.2-0.7).
• The developed statistical framework enhances the ability to identify regulatory genetic variations from AE imbalance.