As trans-eQTLs implicate (direct or indirect) biological relationships, it is somewhat disappointing that none were clearly identified. An alternative way to identify biological relationships is by using co-expression analysis (although such an analysis does not have the aim to identify or explain trans-eQTLs). We conducted such a co-expression analysis and compared it to a co-expression analysis that also takes the genetic variation into account. First, co-expression was assessed between each of the probes that constitute cis-eQTLs (SNP-probe midpoint distance < 250 kb, FDR = 0.05) and all other probes (Figure 4a) that do not map to the same chromosome, by calculating all pair-wise Spearman correlation coefficients. This resulted in 13,003,505 and 42,270,807 pair-wise tests for the celiac peripheral blood dataset (using 765 different cis-probes) and the HapMap B cell line dataset (using 994 different cis-probes), respectively. Through Bonferroni correction we accounted for multiple testing, resulting in the identification of 50,821 and 168,292 significantly correlated transcript-pairs respectively (Spearman's correlation test P-Value < 0.05 after multiple testing correction).
