Comprehensive expression-based isoform biomarkers predictive of drug responses based on isoform co-expression networks and clinical data.

Drug sensitivity biomarkers are molecular features informative for drug response. Previous studies have identified many candidate drug-sensitivity signatures at the transcript level based on significant p values. However, the potential of sensitivity biomarkers has not been sufficiently understood because these investigations have focused on individual biomarkers and have not been carried out at the systems level. In this study, we applied a meta-analytical framework to compute co-expression between isoform pairs in two large datasets of RNA-seq profiles of breast cancer cell lines. We then built hallmark-related direct (HRD) networks by integrating a breast cancer specific isoform co-expression (BCIC) network and hallmark-related isoforms. Next, we explored the associations between isoform biomarkers and the functional clusters of the HRD network. The crucial isoform-based biomarkers for drugs were identified by functional clusters analysis and elucidated by combining isoform expression profiles with clinical information for breast cancer in The Cancer Genome Atlas.