Fig. A.8.

This figure depicts the true signal-to-noise ratio for each iteration of iRF-LOOP for the DREAM challenges as the blue line. The orange line is the corresponding GENIE3 network thresholded to match the same number of edges as the iRF-LOOP network, used to confirm that the unsupervised thresholding contributes to the improvement in signal-to-noise ratio. Unlike the larger empirical networks shown in Fig. 3, these DREAM networks are too small to make a considerable difference when thresholdi.ng the networks.