Application of a Genomic/Transcriptomic Deconvolution Toolbox Addressing Tumor Heterogeneity of Consensus Molecular Subtypes in Metastatic and Primary Colorectal Cancer Samples

In colorectal cancer (CRC), confounding due to cellular heterogeneity within and between samples contributes to variations in molecular characterization of the tumor and consequently in clinical outcomes of cancer. Despite increasing interest in addressing the problem of cellular heterogeneity, approaches for integrating high-throughput sequencing data from mixed tissues are still inadequate. Our primary objective is to improve CMS classifications of heterogeneous metastases and primary colorectal tumor samples through transcriptomic deconvolution. We will develop a colorectal cancer specific deconvolution toolbox for metastases and primary CRC samples to improve CMS classification. We will also apply the deconvolution toolbox to deconvolute immune subtypes of metastases and primary colorectal cancer data.