Integrative eQTL analysis of tumor and host omics data in individuals with bladder cancer

ABSTRACT Integrative analyses of several omics data are emerging. The data are usually generated from the same source material (i.e., tumor sample) representing one level of regulation. However, integrating different regulatory levels (i.e., blood) with those from tumor may also reveal important knowledge about the human genetic architecture. To model this multilevel structure, an integrative‐expression quantitative trait loci (eQTL) analysis applying two‐stage regression (2SR) was proposed. This approach first regressed tumor gene expression levels with tumor markers and the adjusted residuals from the previous model were then regressed with the germline genotypes measured in blood. Previously, we demonstrated that penalized regression methods in combination with a permutation‐based MaxT method (Global‐LASSO) is a promising tool to fix some of the challenges that high‐throughput omics data analysis imposes. Here, we assessed whether Global‐LASSO can also be applied when tumor and blood omics data are integrated. We further compared our strategy with two 2SR‐approaches, one using multiple linear regression (2SR‐MLR) and other using LASSO (2SR‐LASSO). We applied the three models to integrate genomic, epigenomic, and transcriptomic data from tumor tissue with blood germline genotypes from 181 individuals with bladder cancer included in the TCGA Consortium. Global‐LASSO provided a larger list of eQTLs than the 2SR methods, identified a previously reported eQT...
Source: Genetic Epidemiology - Category: Epidemiology Authors: Tags: BRIEF REPORT Source Type: research