Resolving clonal substructure from single cell genomic data in primary and metastatic tumors using CopyKit

Advances in single cell copy number sequencing technologies have enabled the generation of data on hundreds to thousands of cells in parallel. Despite the rapid development in these technologies, there is a significant bottleneck for the analysis of the resulting large-scale datasets. Here we present CopyKit, a comprehensive and user-friendly toolkit for the analysis of single cell copy number data. CopyKit provides a suite of tools for pre-processing, QC, copy number inference, and subclone clustering to delineate the clonal diversity of tumors. We performed single cell copy number sequencing of 2977 cells from two patients breast tumor liver metastasis. CopyKit identified 4 and 12 subclonal populations, including amplification of PDGFRA, KRAS, and MYC as well as losses of PTEN, FOXO1, and RB1, many of which were spatially segregated in the tumor mass of the two tumors. Additionally, we applied CopyKit to study metastatic dissemination of 391 cells from a primary ER+ breast tumor with two matched metastatic sites and 4451 cells from two colorectal carcinomas with matched liver metastatic tissues. This analysis uncovered that the metastatic samples from the breast tumor lacked the ERBB2 amplification present in the primary site. The liver metastasis had deletions in chromosomes 18, 19, and 20p and a focal gain of FGFR1, while the pleural effusion acquired two additional focal gains on chromosome 8, including MYC and BRAF. The colorectal metastatic tumors diverged from the primary tumor with copy number events affecting important cancer genes such as gain of SOX4, MYC, CDK8 and deletions of FHIT, CHEK1, and CHEK2. Collectively, this study shows that CopyKit provides a comprehensive set of tools for resolving clonal substructure from single cell copy number data for diverse applications in cancer biology