(Proceedings of the National Academy of Sciences, 2015) Jiang, Peng; Freedman, Matthew; Liu, Jun; Liu, Xiaole
We developed an efficient and accurate computational framework, RABIT (regression analysis with background integration), and comprehensively integrated public transcription factor (TF)-binding profiles with TCGA tumor-profiling datasets in 18 cancer types. To systematically search for cancer-associated TFs, RABIT controls the effect of tumor-confounding factors on transcriptional regulation, such as copy number alteration, DNA methylation, and TF somatic mutation. Our predicted TF regulatory activity in tumors is highly consistent with the knowledge from cancer gene databases and reveals many previously unidentified cancer-associated TFs. We also analyzed RNA-binding protein regulation in cancer and demonstrated that RABIT is a general platform for predicting oncogenic gene expression regulators.
