Fig. 3
Comparisons of TmGRNs, TTGRNs and mmGRNs across different scenarios. A Overlap of mmGRNs, TmGRNs, and TTGRNs from mean data for the top 1 million edges. B Overlap between GRNs and ChIP-seq regulatory networks. The control consisting of TF-target pairs was generated randomly using the same TF sets and target sets for each dataset. The random TF-target pairs were generated 1,000 times for each GRN in each dataset. The average value of the overlaps between random targets and ChIP-seq targets was compared to the overlaps between GRN predictions and ChIP-seq targets using the χ2 test. In addition, because three types of GRN were constructed using the same TFs set and targets set in each dataset, the comparisons of overlapping targets among mmGRN, TTGRN, and TmGRN were also performed using the χ2 test. C The overlapped target number for each TF in 3 types of GRNs from both mean and bulked data (Student paired t test). D The overlaps between GRN and ChIP-seq for each TF after normalization by GRN size of each TF (Student paired t test). E The overlaps between GRN and ChIP-seq for each TF after normalization by ChIP-seq target number (Student paired t test). F, G Target comparisons between different GRNs and ChIP-seq for FEA4 (F) and KN1 (G). Top 100, top 200, and top 500 targets were considered. Significant differences were determined using the paired Student t test. H Simple Enrichment Analysis (SEA) [24] for the 1-kb sequences upstream for the start codons of the top 100 targets of 11 TFs in three types of GRNs, the center line in box represents the median. TP: Number of primary sequences matching the motif / number of primary sequences (percentage of primary sequences matching the motif); primary sequences are from the predicted targets. FP: Number of control sequences matching the motif / number of control sequences (percentage of control sequences matching the motif); control sequences are from the genes that randomly selected in the set of expressed genes. I Comparisons of GRN weights of three types of GRNs at different thresholds, significant differences were determined using the paired Student t test, center line in box represents the median. In all tests of significance above, “*” represents P < 0.05; “**” represents P < 0.01; “***” represents P < 0.001