Computational identification of genetic polymorphisms influencing human gene expression in obesity gene

Abstract

The heritability of obesity has been estimated to be between 40% and 70%. 60 GWAS and more than 1,100 loci were reported. Most of these loci are in non-coding regions, making it more difficult to understand the role of these variants in the disease. One of the methods to understand these non-coding variants is to estimate their effects on gene expression levels of the neighbouring gene (cis-eQTL) or far away genes (trans-eQTL). This is achieved by a regression model explaining the gene expression level by the genetic variant and other covariates. The GTEx project characterized genetic effects on transcriptome across different tissues with eQTLs but did not report any eQTL on the principally expressed tissue of genes involved in obesity.

Our project employed a rigorous eQTL mapping approach, utilizing gene expression and whole genome data from the reputable GTEx database. The genotype data, obtained from the GTEx consortium, was meticulously divided by each of the 22 chromosomes. The expression data, downloaded from the GTEx portal, was carefully processed into an expression matrix. Covariates were included to adjust for principal components, sex, PEER factors and protocol. The MatrixeQTL model, a well-established method, was used for the eQTL mapping of 21 genes related to the leptin-melanocortin pathway in tissues where these genes are highly expressed (pituitary gland, hypothalamus, and adipose visceral tissue). Our thorough approach led to the identification of 8221 eQTLs, with the gene POMC having the most eQTLs. This project generated a set of cis—and trans-eQTLs. These eQTLs may explain the variability of gene expression in genes related to obesity. They can be used for follow-up analyses, including colocalization or Mendelian randomization, to highlight the effect of these variants directly on the obesity phenotype.