Accurate prediction and elucidation of drug resistance based on the robust and reproducible chemoresponse communities

Selecting the available treatment for each cancer patient from genomic context is a core goal of precision medicine, but innovative approaches with mechanism interpretation and improved performance are still highly needed. Through utilizing in vitro chemotherapy response data coupled with gene and miRNA expression profiles, we applied a network‐based approach that identified markers not as individual molecules but as functional groups extracted from the integrated transcription factor and miRNA regulatory network. Based on the identified chemoresponse communities, the predictors of drug resistance achieved high accuracy in cross‐validation and were more robust and reproducible than conventional single‐molecule markers. Meanwhile, as candidate communities not only enriched abundant cellular process but also covered a variety of drug enzymes, transporters, and targets, these resulting chemoresponse communities furnished novel models to interpret multiple kinds of potential regulatory mechanism, such as dysregulation of cancer cell apoptosis or disturbance of drug metabolism. Moreover, compounds were linked based on the enrichment of their common chemoresponse communities to uncover undetected response patterns and possible multidrug resistance phenotype. Finally, an omnibus repository named ChemoCommunity (http://www.jianglab.cn/ChemoCommunity/) was constructed, which furnished a user‐friendly interface for a convenient retrieval of the detailed information on chemoresponse communities. Taken together, our method, and the accompanying database, improved the performance of classifiers for drug resistance and provided novel model to uncover the possible regulatory mechanism of individual response to drug. What's new? Selecting the best available treatment for individual cancer patients based on their genomic makeup is a core goal of precision medicine, but better approaches are still needed. Using a network‐based approach and in vitro chemotherapy response data coupled with gene and miRNA expression profiles, here the authors identified markers not as individual molecules but as functional groups extracted from the integrated transcription factor and miRNA regulatory network. The network‐based algorithm and accompanying database may help identify novel types of drug resistance markers with better predictive performance and improve the understanding of transcriptional and post‐transcriptional regulation mechanisms involved in drug resistance.