Initial data mining of results from several on-going public cancer genomics initiatives has identified both known and novel somatic alterations across multiple cancer types. A challenge of validating putative cancer targets discovered by such studies is the identification of appropriate preclinical cell line models with relevant genetic alterations. Thus there is an urgent need for comprehensive mutation information across a large panel of cancer cell lines. Mutation discovery in cell lines brings its own unique challenges. Most cancer re-sequencing studies published to date typically use DNA derived from both primary tissue and matched DNA from a non-cancerous sample, often from peripheral blood or adjacent tissue. Cell lines present a challenge for discovering somatic mutations because a matched normal sample that can be used to filter for somatic mutations rarely exists. To this end we have implemented a filtering scheme that attempts to enrich for true-somatic mutations. To assess the effectiveness of the filtering pipeline, we compared filtered results with mutations discovered in matched normal cell lines derived from blood for 10 cancer cell lines. We estimate that filtering captures 68% of true somatic mutations. This rate is slightly improved or comparable over existing methods used in previous studies involving sequencing of cancer cell lines. Interestingly, we find that there is typically 10% of known true somatic mutations that are unable to be called by any of the mutation callers used in a “no-normal” context but are discoverable with a normal present. In such cases, the presence of the normal can add additional information that aids mutation calling. To genetically characterize a panel of cell lines for pharmacological studies, we performed exome sequencing of 223 cancer cell lines. Analysis of filtered mutation variants reveals that cell lines harbor an average of 270 nonsynonymous coding variants, an increase over typical amounts found in primary tumors. Furthermore we find that for several tumor lineages, the spectra of mutations observed reflect the mutation signatures identified by other large scale efforts in primary tumors. With regard to recurrently mutated oncogenes and tumor suppressors, we observe significant overlap with existing mutation data derived from primary tumor samples. Additionally, we identify relevant cell line models for several novel cancer driver genes reported in recent studies. As a resource for the scientific community, we have made both aligned read data as well as mutation calls in various formats freely available and easily accessible.
Citation Format: Alex H. Ramos, Ruibang Luo, Jacob Feala, Binghang Liu, Lara Gong, Markus Warmuth, Ping Zhu, Peter Smith, Lihua Yu. Exome sequencing of tumor cell lines: Optimizing for cancer variants. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4269. doi:10.1158/1538-7445.AM2014-4269