Computational inference of co-regulatory modules from transcription factors, MicroRNAs, and their targets using CanMod2.

Gene regulators such as Transcription Factors (TFs) and microRNAs (miRNAs) regulate genes at the transcriptional and post-transcriptional levels, respectively. There is a complex interplay of regulatory patterns of TFs and miRNAs. Some TFs and miRNAs regulate the activity of their target genes individually, some co-regulate the activity of the same set of genes, some TFs regulate miRNA activity, and some miRNAs regulate TFs. As dysregulation in gene regulation can lead to various diseases like cancer, it is a significant problem to find the interplay among TFs, miRNAs, and their target genes. Here, we propose a computational pipeline, CanMod2, which infers modules of TFs, miRNAs, and their co-regulatory targets that are involved in a common biological process. In this work, we have introduced several algorithmic enhancements to the earlier version of CanMod2. We applied CanMod2 to five cancer types and analyzed the inferred modules extensively. Our results show that the inferred modules were enriched in cancer-related biological processes and pathways. The hub regulators that occur in many modules were among cancer-related genes and miRNAs. The inferred regulator-target interactions were significantly enriched in ground truth interactions. CanMod2 source code and documentation are publicly available at https://github.com/bozdaglab/CanMod2 .