Theoretical framework for <i>cis</i>-mediation of <i>trans</i>-eQTLs.

<p>Panel (A) shows a causal diagram, in which a causal variant (SNP_causal) affects expression of a <i>cis</i>-transcript (gene<i>_cis</i>) which in turn affects expression of a distant gene (gene<i>_trans</i>). When SNP_causal is measured and is the strongest associated SNP (i.e., “lead SNP”) for both the <i>trans</i>- and <i>cis</i>-eQTL association signals, no other SNPs are involved in mediation analysis. Panel (B) shows the causal diagram underlying an eQTL mediation analysis when the causal variant is unmeasured or is not the lead SNP for both the <i>trans</i>- and <i>cis</i>-eQTL association signals. Thus, the lead SNP for gene<i>_cis</i> and gene<i>_trans</i> may be different, and are noted here as SNP_{lead-<i>cis</i>} and SNP_{lead-<i>trans</i>}, respectively. Solid lines represent causal effects, and dotted lines represent non-causal correlation, including linkage disequilibrium (LD).