Using GCPBayes to explore pleiotropy at gene-level between breast and ovarian cancers based on GWAS summary statistics data

Background/Objectives: Cross-phenotype association at gene- or pathway-level can help to detect pleiotropic genes, i.e. the fact that one gene can affect multiple traits, and inform about common mechanisms between diseases. However, the lack of proper pipelines to apply gene-set analysis in this context using GWAS data in a reasonable running time could prevent researchers to apply such methods.

Methods: We designed a user-friendly pipeline to perform cross phenotype association using GCPBayes method [1], a cross-phenotype gene-based method developed by our team. We illustrated the application on publicly available GWAS summary statistics on breast cancer (BC) and ovarian cancer (OC) from BCAC and OCAC consortia and compared the results with previous studies used SNP-level analysis or transcriptome data.

Results: Previous studies suggested 40 pleiotropic genes but only one (RCCD1) was replicated by two studies. Our method retrieved seven of these genes: TERT, BABAM1, CPNE1, RGS19, SMC2, CLIC6, and RCCD1. Besides, we also detected additional 140 new genes with potential pleiotropic signals for BC and OC. However, we are working on a suitable way to narrow our large list of candidate genes for further experimental analyses.

Conclusion: Our method replicated some genes previously found associated to both BC and OC using gene-set rather than SNP-level approach. User-friendly tutorials are available on our group’s GitHub page.