ABSTRACT
Colorectal cancer (CRC) is a common cancer with a high mortality rate and a rising incidence rate in the developed world. The disease shows variable drug response and outcome. Molecular profiling techniques have been used to better understand the variability between tumours as well as cancer models such as cell lines. Drug discovery programs use cell lines as a proxy for human cancers to characterize their molecular makeup and drug response, identify relevant indications and discover biomarkers. In order to maximize the translatability and the clinical relevance of in vitro studies, selection of optimal cancer models is imperative. We have developed a deep learning based method to measure the similarity between CRC tumors and other tumors or disease models such as cancer cell lines. Our method efficiently leverages multi-omics data sets containing copy number alterations, gene expression and point mutations, and learns latent factors that describe the data in lower dimension. These latent factors represent the patterns across gene expression, copy number, and mutational profiles which are clinically relevant and explain the variability of molecular profiles across tumours and cell lines. Using these, we propose a refined colorectal cancer sample classification and provide best-matching cell lines in terms of multi-omics for the different subtypes. These findings are relevant for patient stratification and selection of cell lines for early stage drug discovery pipelines, biomarker discovery, and target identification.
