Figure 6.

Two left panels represent the results of the analysis of the simulated eQTL data with the conditional single marker tests based on the model (18) (left) and the stepwise selection strategy based on the model (3) (middle). Values on the y-axis and x-axis correspond to the indices of the e-traits and the markers, respectively. Red points mark the positions which are significant at the Genome Wise Type I Error Rate (GWER) of 0.05/1500. In the last panel, we present means of the absolute values of t-test statistics over all 1500 traits (black line) and the theoretical expected values calculated according to the formula (19) in the Appendix (red line). In most positions, these curves coincide completely. The blue line represents 75% quantiles of the t-test statistics absolute values distribution along the genome. The horizontal lines correspond to the Genome Wide adjusted thresholds of Dupuis and Siegmund (1999) for GWER control at the levels 0.05 and $0.05/1500$. The larger threshold uses the Bonferroni correction to adjust to the number of traits.