CATS—A tool to contextualize cancer hallmarks to specific cancer types

Gene signatures, widely used to infer the activity of biological processes based on transcriptomic data, lack tissue/cellular context. For instance, the epithelial-mesenchymal transition (EMT) signature derived from one cellular context may not be precisely applicable to another context; overlapping yet distinct genesets are likely to mediate EMT in different contexts. Here, we derive cancer type-specific gene signatures for 14 oncogenic hallmark processes across 23 cancer types by integrating the context-agnostic reference gene signatures with the protein interaction network and cancer-specific transcriptomic data. Overall, our inferred context-specific gene signatures exhibit a higher cancer specificity than the corresponding reference genesets with respect to a variety of biological and clinical features, such as activity in malignant cells, genetic vulnerabilities, patient prognosis, and treatment response. We provide the projections for 14 cancer hallmarks across 23 human cancer types, along with the CATS (Cancer Specific Transcriptomic Signature) software tool and the evaluation repository.