Cross-GWAS coherence test at the gene and pathway level

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Genome-wide association studies (GWAS) often detect correlated genetic variants, leading to non-independent effect sizes.
• Existing methods for aggregating GWAS effects are limited to single studies across genes or pathways.

Purpose of the Study:

• To develop a robust and efficient method for detecting coherent genetic association signals between two traits across different GWAS.
• To facilitate cross-GWAS analyses by accounting for known inter-sample covariance structures.

Main Methods:

• Devised a new significance test for covariance of dependent data with known inter-sample covariance.
• Utilized a distribution of test statistics as a linear combination of chi-squared distributions.
• Applied Davies' algorithm for precise calculation of the cumulative distribution function.
• Extended the test to identify gene-wise causal links.

Main Results:

• Successfully applied the framework to test for dependence between SNP-wise effect sizes of two GWAS at the gene level.
• Identified potential shared genetic links between COVID-19 severity and rheumatoid arthritis, vitamin D levels, serum calcium, and specific medications (M05B).
• Detected the involvement of chemokine receptor genes and integrin beta-1 related genes in COVID-19 severity.

Conclusions:

• The developed method provides a powerful tool for cross-GWAS analyses, enabling the detection of shared genetic architectures between traits.
• The findings highlight potential genetic pathways influencing COVID-19 severity, offering new avenues for research and therapeutic strategies.
• This approach is broadly applicable to datasets with known auto-correlation structures.