For the 228 individuals for whom we had expression data for all four conditions, we observed 21,516 independent cis SNP-probe associations across data sets (defined as the most significantly associated SNP with expression for a given probe per data set), 24.6% of which had the lowest P value for association in naïve cells, 21.6% after 2-hour LPS, 25.4% after 24-hour LPS, and 28.3% after IFN-γ. A total of 11,476 probes had at least one eQTL, of which 43.4% (4977 of 11,476) only had eQTL after monocyte stimulation. We find that eQTL are similarly specific to the unstimulated state, with 54.1% of naïve eQTL (2866 of 5299) not found after treatment (Fig. 1A, fig. S3, and table S2). One explanation for many context-specific eQTL is that treatment increases expression to quantifiable levels, thus allowing eQTL detection. To investigate the extent to which this contributed to the observed stimulus specificity of eQTL, we restricted our analysis to probes with detectable expression (Illumina detection score P < 0.01) in more than 90% of samples from each data set. We found that 33.4% (1702
