Proteogenomics of Cancer: Peptides to Events

Cancer is marked by a progression of somatically acquired genomic lesions. Recent technological advance help the researcher to have cheaper genomic data which led us to a better understanding of the mutations that drive the progression of diseases. However, the impact of mutations at the protein level have not been as well understood.

Using MS-based proteogenomics, many of these mutations can be confirmed directly at the protein level. The challenging problem is how to integrate the huge amount of genomic data into comprehensive, yet compact, protein databases, and how to analyze peptide result systematically while the peptide sequence itself doesn't include enough information.

I introduce the tools 'SpliceDB' and 'Enosi'. SpliceDB uses a graph based approach to model junction and variant peptides, and deriving compact linear databases from them. From the databases deriven from SpliceDB, Enosi proposes a novel FDR strategy to improve sensitivity and combine peptide evidence to 'events' which categorized by different type of genomic variation. We reanalyzed the colorectal cancer data on 90 individuals from the CPTAC project. From 2,367 novel peptide identifications, we reported 1,884 proteogenomic events.