Degree Centrality of SNPs in EQTL Networks

• Activity Number: 530 - Integrative Genomics: EQTL and GWAS
• Type: Contributed
• Date/Time: Wednesday, August 1, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Sheila Gaynor* and Maud Fagny and John Platig and Xihong Lin and John Quackenbush
• Companies: Harvard University and Dana Farber Cancer Institute and Dana Farber Cancer Institute and Harvard University and Dana Farber Cancer Institute
• Keywords: eQTL; networks; GTEx; marginal models; two part model; degree centrality

Abstract

Network analyses are a natural approach for identifying genetic variants and genes that work together to drive disease phenotype. The relationship between SNPs and genes, captured in expression quantitative trait locus (eQTL) analysis, can be represented as a network with edges connecting SNPs and genes. We propose alternative degree metrics to represent how central and potentially influential a SNP is to an eQTL network, and estimate them as functions of the eQTL regressions. We apply our metrics to data from the GTEx project to assess whether SNPs strongly associated to particular diseases are more central to disease-specific tissues. We characterize features of the proposed metrics, including how well they replicate. We further introduce novel marginal two part models to assess whether SNPs associated with esophagus cancer and type 2 diabetes are more central to eQTL networks in esophageal and adipose tissue, respectively.