Processed gene expression data from the published Merck liver eQTL study [31] were downloaded from GEO (GSE9588, GPL4371). Based on available sample metadata, 266 samples had (a) unambiguous sample ID, age and sex assignments (b) expression data, (c) genotype data, and (d) did not overlap with the UC study. Probes were grouped into RefSeq gene annotation probe sets based on the array manifest. Probesets were further clustered and split following the methodology used for the UC array set.
