Abstract:
|
Copy number alterations (CNAs) are large-scale, somatically derived structural genomic changes implicated in formation and progression of multiple cancers. While a number of methods exist to determine, for a given patient, regions of CNA, there is substantial need for methods to compare profiles across patient subgroups. In this work, we demonstrate a method for group comparison that relies on Bayesian change-point models for CN profile characterization, coupled with a clustering enabled inference on subgroup comparison. We illustrate the method using publicly available data from the Cancer Cell Line Encyclopedia (CCLE). CCLE is a set of 947 profiled human cancer cell lines annotated for multiple genetic and pharmacologic information, with copy number measurements from the Affmetrix SNP6.0 platform. We examine the ability of our inferential method to compare copy number patterns across cancer type, and to identify subgroups of drug response categories available for a subset of the CCLE data. A performance evaluation using simulated copy number profiles is given.
|