Fig. 5

HiC data is capable of capturing the effect of long distance interactions between TF binding within distal TFBS and gene’s promoter on gene expression. (a) We used the cell line specific Hi-C data to weight the distal TF-TG interactions in the motif adjacency matrix. We also down-weighted or up-weighted the interactions with the promoter TFs which would have been missed otherwise due to the low resolution nature of Hi-C data. (b) We predicted expression of GM12878 and K562 TGs using distal TFBS based GRNs with and without HiC data integration in order to evaluate its predictive value for the models. C shows the predictive performance of the models using GRNs containing HiC normalized motif edges based on the Hi-C UP weighting scheme compared to those built using unweighted binary motif network without HiC information