Abstract 3959: COSMIC: Mining the world's knowledge on somatic cancer mutations

Abstract

Abstract COSMIC, the Catalog Of Somatic Mutations In Cancer (www.sanger.ac.uk/cosmic) is a single resource combining much of the available data on somatic mutations in human cancer. This information is obtained from a number of online sources, including the TCGA and ICGC webportals, the IARC p53 database and the Cancer Genome Project, Sanger Institute, UK; this is then combined with information carefully curated from the scientific literature. Over 20,000 genes have been screened in the search for cancer causing mutations, and details have been curated for all of these, resulting in the annotation of 213,615 mutations across 665,763 tumor samples (COSMIC v56, Nov 2011). The Cancer Gene Census lists all known cancer genes and these are prioritized for curation, generating extensive mutation spectra across 107 point-mutated genes and 111 fusion gene pairs. Increasing whole genome/exome data are also available; currently 559 samples detail 2753 genome rearrangements, 21892 coding and 57396 non-coding point mutations. Most importantly, this information is all rigorously maintained and regularly updated. With new releases every two months, COSMIC is very responsive to the scientific literature, detailing all the latest findings. The data in COSMIC can be examined in a number of different ways. The website has been designed to allow easy graphical browsing of the data, with many mining options to filter desired information. A COSMIC genome browser can be used to visualize COSMIC mutations in a genomic context, and a Biomart is available to download datasets federated with external databases. In addition, the complete data is available for free download for offline analysis. The combination of data from multiple sources in COSMIC allows its use as a discovery tool as well as a reference datasource; substantial mining can be performed between mutant genotypes and clinical phenotypes, searching for novel recurrence patterns between tissues, genes and individual variants. Overall, the top mutated gene reported in human cancer is JAK2, with a published mutation rate of 38% (n=76853), however a phenotype correlation suggests a mutation rate of up to 86% in Polycythaemia Vera, 58% in Myelofibrosis and 55% in Essential Thrombocythaemia. Similarly, KRAS with an overall mutation rate of 23% (n=96055) increases to 69% in Pancreatic ductal carcinomas, and a single report suggests a 70% mutation frequency for MED12 in Uterine Leiomyomas (n=225). In addition to these well characterized cancer genes, new whole-genome screens are beginning to identify many more. For instance the TCGA Ovarian exome screen suggested roles for genes such as CSMD3, HMCN1 and USH2A, and combining multiple whole genome screens suggests roles for further new genes such as BAI3 (13% mutated in lung). As yet, novel recurrent mutations are challenging to identify across studies, but rapidly increasing amounts of whole genome data will make COSMIC a key resource in this search for new therapeutic targets. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3959. doi:1538-7445.AM2012-3959