| dc.description.abstract | Cancer is a heterogeneous, complex disease and one of the leading causes of death worldwide. The high heterogeneity between individuals, even those affected by the same cancer type, can lead to distinct molecular and phenotypic tumor profiles that affect their response to treatment. In silico modeling of cancer as a system of interacting molecules can provide the basis to enhance biological understanding about disease progression, and the means to tailor treatment to specific groups of patients or even individuals based on their molecular alterations. Here we investigate how patients omics data can provide insights regarding dysregulated processes in the Consensus Molecular Subtypes of colorectal cancer and how the modeling of these processes as functionally-related groups of entities can provide a more intuitive and manageable level of complexity to analyze, understand and use biological models. We found that those groups of entities, called modules, have distinct associated functions that can promote or suppress cancer development, and can be linked to clinical practice through their association with patient survival and the development of targeted therapies. Furthermore, the modules were used as "building blocks" for the extension of an existing cell-fate decision model, to better capture colorectal cancer-specific processes. Prediction of drug effects using the extended model revealed that the model had a higher predictive ability for adenocarcinoma cell lines in general, indicating that optimizing the model for colorectal cancer improved the representation of cancer types sharing the same histology. Our results demonstrate how a modular approach to modeling a biological system can benefit in silico experimentation in cancer, helping to bring closer patient-specific molecular alterations and treatment decisions in the clinic. | |