While these early studies have emerged as proof of principle that novel SNVs can indeed be discovered using NGS, the study of computational methods for their discovery in cancer is under-represented in the bioinformatics literature. The analysis of SNVs from cancer data, where altered ploidy and tumor cellularity impact the statistical expectations of SNV discovery; and transcriptome data, where the dynamic range of depth of sequencing is dependent on highly variable transcript expression present unique challenges. In this contribution, we describe a new statistical model for identifying SNVs in NGS data generated from cancer genomes and transcriptomes. We demonstrate how its novel features outperform other available methods. Additionally, we provide a ground truth dataset (with Sanger validated SNVs) and robust accuracy metrics that will permit future study of computational methods for SNV detection in cancer genomes.
