Living cells are realized by complex gene expression programs that are moderated by regulatory proteins called transcription factors (TFs). The TFs control the differential expression of target genes in the context of transcriptional regulatory networks (TRNs), either individually or in groups. Deciphering the mechanisms of how the TFs control the expression of target genes is a challenging task, especially when multiple TFs collaboratively participate in the transcriptional regulation. Recent developments in biotechnology have been applied to uncover TF-target binding relationships to reconstruct draft regulatory circuits at a systems level. Furthermore, to identify regulatory interactions in vivo and consequently reveal their functions, TF single/double knockouts and over-expression experiments have been systematically carried out. However, the results of many single or even double-knockout experiments are often non-conclusive, since many genes are regulated by multiple TFs with complementary functions. To predict the TF combinations that the knocking out of them are most likely to bring about the phenotypic change, we developed a new computational tool called TRIM that models the interactions between the TFs and the target genes in terms of both the TF-target interaction's function (activation or repression) and its corresponding logical role (necessary and/or sufficient). We used DNA-protein binding and gene expression data to construct regulatory modules for inferring the transcriptional regulatory interaction models for the TFs and their corresponding target genes. Our TRIM algorithm is based on an HMM and a set of constraints that relate gene expression patterns to regulatory interaction models. However, TRIM infers up to 2-TFs interactions. Inferring the collaborative interactions of multiple TFs is a computationally difficult task, because when multiple TFs simultaneously or sequentially control their target genes, a single gene responds to merged inputs, resulting in complex gene expression patterns. We developed mTRIM to solve this problem with a modified association rule mining approach. mTRIM is a novel method to infer TF collaborationsin transcriptional regulation networks. It can not only identify TF groups that regulate thecommon targets collaboratively but also TFs with complementary functions. However, mTRIM ignores the effect of miRNAs on target genes. In order to take miRNAs' effect into considerations, we developed a new computational model called TmiRNA that incorporates miRNAs into the inference. TmiRNA infers the interactions between a set of regulators including both TFs and miRNAs and the set of their target genes. We used our model to study the combinatorial code of Human Cancer transcriptional regulation.
Read
