Bayesian test for colocalisation between pairs of genetic association studies using summary statistics

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Complex Disease Aetiology

Background:

• Genome-wide association studies (GWAS) have identified numerous genetic associations with human diseases and traits, particularly cardiovascular diseases and lipid biomarkers.
• Understanding the molecular basis of these GWAS associations remains a significant challenge.
• Integrating diverse association datasets, including gene expression data, can help elucidate these molecular mechanisms.

Purpose of the Study:

• To develop and validate a novel statistical methodology for assessing shared causal variants between different genetic association signals.
• To integrate genome-wide association study (GWAS) data with expression quantitative trait locus (eQTL) studies to identify candidate causal genes.
• To provide a framework for systematic meta-analysis comparisons across multiple GWAS datasets.

Main Methods:

• Developed a novel statistical method to test for consistency between two association signals, indicative of a shared causal variant.
• Applied the method to integrate a gene expression dataset (966 liver samples) with a large meta-analysis of lipid traits (>100,000 individuals).
• Utilized single nucleotide polymorphism (SNP) summary statistics to enable broad applicability and meta-analysis comparisons.

Main Results:

• The re-analysis supported 26 out of 38 previously reported colocalisation results between lipid traits and eQTLs.
• Identified 14 novel colocalisation results, demonstrating the value of the formal statistical testing approach.
• Re-assigned causality for three eQTL-lipid pairs, identifying alternative candidate causal genes (SORT1, GCKR, KPNB1) as more likely.

Conclusions:

• The novel statistical methodology effectively integrates GWAS and eQTL data to identify shared causal variants.
• The approach enhances the discovery of candidate causal genes underlying complex disease associations.
• This methodology has significant implications for understanding complex diseases and designing targeted drug therapies.