All tested gene-SNP pairs and significant eQTL pairs in the GTEx liver cohort (version 7) were downloaded from the GTEx portal. The GTEx dataset includes 153 liver samples with available genotype and gene expression data of which only 15 of African descent (based on phenotype file, verified by PCA, S5 Fig). The GTEx dataset was filtered by SNPs with minor allele frequency (MAF)>0.05, leaving 6,453,712 SNPs. A total of 311,967 eQTLs and 3,416 eGenes in the GTEx liver were retained with this filtering.
We overlapped the AA hepatocyte eQTLs with the GTEx eQTLs to identify overlapping eQTLs and AA-specific eQTLs. AA hepatocyte eQTLs that were also significant in the GTEx dataset and have concordant direction of effect were identified as overlapping eQTLs. AA hepatocyte eQTLs that were not found in GTEx were identified as AA-specific eQTLs.
To compare AA hepatocyte eQTLs with GTEx liver eQTLs in the European background, we selected 127 samples that are of European ancestry from GTEx liver samples (S5 Fig) and performed eQTL mapping with FastQTL [76] as described in GTEx v7. We performed 1000 permutation to identify the threshold of eGene discovery and selected corresponding eQTLs at FDR<0.05. We found a similar pattern of overlapping results as the results using all samples. (Fig 1B and S3 Fig). We also randomly sampled 60 samples from GTEx European liver samples to match the sample size of AA hepatocytes and performed eQTL mapping as described above (total of 10 iterations). We identified the number of overlapping and unique eQTLs/eGenes averaged across 10 iterations.
We used CAVIAR (CAusal Variants Identification in Associated Regions) [33] to statistically fine-map eQTLs in AA hepatocytes and GTEx subsampled eQTLs separately. We applied CAVIAR to the cis eQTL mapping z scores and LD for each overlapping eGene while setting the number of causal variants to 1. CAVIAR estimates the set of SNPs that account for 95% of the posterior probability of the causal variants at each locus.
Do you have any questions about this protocol?
Post your question to gather feedback from the community. We will also invite the authors of this article to respond.