Item

Abstract

Several studies have investigated the effects of genetic variation on gene ex- pression (expression quantitative trait loci, eQTLs) in peripheral tissue, cell lines, or post-mortem brain tissue. EQTL studies from pre-mortem, fresh- frozen brain samples would be highly interesting but are hampered by the restricted accessibility of such samples. At the University of Bonn, we have access to a unique sample of pre-mortem human hippocampus samples originating from surgery of treatment-resistant epilepsy patients. To sy- stematically determine eQTLs in a total of 148 hippocampus samples, we generated whole-genome SNP (Illumina Human660W) and gene expression data (Illumina HumanHT-12v3). In addition to the conventional data ana- lysis, we applied a new “hidden factor” analysis that identifies and corrects for unknown confounding factors in the data and thus diminishes the false- positive and false-negative eQTL rate (PEER, https://github.com/PMBio/ peer/wiki). Fifteen hidden factors were identified and used as co-variates for expression analysis. We detected 78 trans-regulating (>1Mb between SNP and probe) eQTLs that withstood Bonferroni correction for multiple testing. Moreover, 1,925 cis-regulating (≤1Mb distance) eQTLs remained significant after permutation-based Westfall-Young correction. In an addi- tional step, we extended our analysis to the systematic investigation of the influence of DNA methylation on gene expression. Genome-wide methyla- tion measurement was performed using Illumina’s new HumanMethylati- on450 array which interrogates more than 485,000 methylation sites. To our knowledge, our study is the first to integrate genotype, expression and methylation data from pre-mortem brain tissue and will provide a valuable resource for the functional interpretation of genetic and epigenetic sites, in particular those associated with brain diseases.