With the goal of understanding complex mechanism of chemoresponse and developing effective predictors of chemoresistance at the transcriptional and post-transcriptional level, we pursued a simple but efficient method for recognizing specific network communities combined with chemotherapy response, gene expression and miRNA expression profiles in the integrated regulatory network. These chemoresponse communities in question are not encoded as individual molecules, but as communities of interacting gene and miRNA within a larger human transcriptional and post-transcriptional network. Our resulting chemoresponse communities embodied many significant biology and pharmacology components, such as CYP3A4, BCL2, MDR1 and has-miR-21, and displayed novel models to interpret chemotherapy response. To improve the performance of chemoresistance predictors, the identified chemoresponse communities were further used as features with several machine learning algorithms to classify the sensitive/resistant samples. Especially, we also connected the experimental compounds based on enrichment of chemoresponse communities to exhibit the compound response patterns and uncover potential multidrug resistance phenotype for linked compounds.
Enyu Dai, Jing Wang, Feng Yang, Xu Zhou, Qian Song, Shuyuan Wang, Xuexin Yu, Dianming Liu, Qian Yang, Hong Dai, Wei jiang*, Hong Ling*. Accurate prediction and elucidation of drug resistance based on the robust and reproducible chemoresponse communities. International Journal of Cancer. 2018 Apr 1. 142(7):1427-1439.