Single cell multi-omic mapping of subclonal architecture and pathway phenotype in primary gastric and metastatic colon cancers

Abstract

ABSTRACTSingle cell genomics provides a high-resolution profile of intratumoral heterogeneity and subclonal structure in primary and metastatic tumors. Notably, metastases and therapeutic resistant tumors often originate from distinct subclones. These distinct cellular populations are an important contributor to adaptation and resistance to ongoing therapy. Single cell DNA-sequencing (scDNA-seq) defines subclones but does not provide biological information about cell types. Single cell RNA-sequencing (scRNA-seq) provides biological information but is less useful for identifying different subclones. The integrated scDNA-seq and scRNA-seq data from the same tumor cell population provides both subclone structure and biology. To understand the cellular genomic variation of different subclones in primary and metastatic cancers, we developed an approach to integrate multi-omics data from both types of single cells. This joint data set represented thousands of normal and tumor cells derived from a set of primary gastric and metastatic colorectal cancers. The extensive cellular sampling provided robust characterization of the subclonal architecture of gastric and colorectal cancers. We reconstructed the subclonal architecture using the cells in G0/G1 phase. The scDNA-seq provided a ground truth for copy number-based subclones. From the scRNA-seq data, the epithelial cells in G0/G1 were identified and assigned to specific subclones by a correlation algorithm based on gene dosage. The inferred CNV profiles from the scRNA-seq epithelial cells were assigned subclones identified from the scDNA-seq. Afterward, we determined the biological pathway activities of specific clones. Overall, integrative multi-omics analysis of single-cell datasets is more informative than any individual genomic modality, provides deep insights into the intratumoral heterogeneity, and reveals subclonal biology.