A statistical framework for pathway-level association analysis using summary statistics from meta-analysis for genome-wide association studies

Authors

• H Zhang
• N Chatterjee
• W Wheeler
• K Yu

Abstract

Pathway-based approaches have been considered as an important alternative to the single-marker test for genome-wide association (GWA) studies. A pathway-based test takes advantage of extra biological knowledge on gene function, and is more powerful in uncovering a group of functional genes with moderate or small effect sizes. We develop a novel statistical framework to extend Adaptive Rank Truncated Product (ARTP) method to use summary statistics to perform pathway analyses, in which an external reference is required to provide linkage disequilibrium information between SNPs. We demonstrate the accuracy of our test based on summary statistics by comparing its p-values with the ones calculated from raw data of the Genetic Epidemiology Research on Aging (GERA) Cohort for type 2 diabetes (T2D). We apply our method to 4551 pathways by using the summary data from the DIAbetes Genetics Replication and Meta-analysis (DIAGRAM) Consortium. The meta-analysis consists of 2.5 million SNPs of 12,171 T2D cases and 56,862 controls from European descent populations. The 503 individuals from European descent populations in 1000 genomes project are used as external reference. We also apply our method to the raw data of GERA cohort as a replication dataset, which consists of 7,638 cases and 54,319 controls from European descent populations. After excluding 170 T2D susceptibility SNPs that have been reported in literature and SNPs within +/- 500kb of susceptibility SNPs, we identify 31 pathways with genome-wide significance (Fisher's combined P < 0.05/4551 = 1.1E-5). Our method is useful in providing insights into the genetic architecture of complex diseases.

Scientific Focus Area: Epidemiology

This page was last updated on Friday, March 26, 2021