TOWARDS ACCURATE COPY NUMBER CALLING IN HIGH PLOIDY TUMORS

Accurately determining the abundance of oncogenes and tumor suppressors is necessary to understand the pathogenesis and treatment of cancer patients. However, low purity, tumor heterogeneity, chromosomal instability, and aneuploidy can confound even the most robust per-gene copy number detection methods. These limitations especially affect high ploidy tumor genomes, common in solid cancers, by restricting the ability to make confident biological and clinical inferences, even with high quality copy number data. Our research focuses on comparing the copy number methods of two algorithms. The first is The Cancer Genome Atlas (TCGA) method, which is based on data from single nucleotide polymorphism arrays. The second is a FACETS-based method (Fraction and Allele Specific Copy Number Estimate from Tumor Sequencing) and utilizes whole exome sequencing data. We hypothesized that purity and ploidy-aware inferences of total and allele-specific integer copy number would lead to more accurate and nuanced clinical insights. Our research identifies sources of discordance, which include differences in segmentation and discretization methods. Based on these findings, we suggest metrics of accuracy that use the intersection of TCGA generated RNA data with validation by the alternate method. From this initial accuracy test, we found that FACETS seems to undercall deletions and TCGA seems to undercall amplifications. In an initial attempt to improve copy number methods, we calculated a new per-gene copy number relative to the rest of the genome by using the FACETS derived total copy number and the average sample ploidy. Our accuracy test suggests that the new copy number improves sensitivity at a cost to specificity. Our research lays the foundation for the improvement of methods to detect copy number in complicated tumors in order to advance patient treatment and hopefully progress towards a cure.