Abstract LB-305: A computational approach to identify recurrent somatic driver events in noncoding regions in human cancers

Abstract

Abstract SNVs in non-coding regions can be relevant for cancer making it promising to search for driver somatic events within non-coding regions. For example, a number of recent studies have described the first known recurrent non-coding event in the promoter region of the TERT gene in melanoma and other cancer types (Horn et al., 2013; Huang et al., 2013; Vinagre et al., 2013). Other candidate regions (PLEKHS1, WDR74 and SDHD) were recently identified using a computational approach for which the code is not available (Weinhold et al., 2014). Currently available approaches require narrowing down the list of genomic regions by filtering by particular categories of genomic elements like promoters, enhancers or DNAse-hypersensitive sites or by proximity to TSS (Fredriksson et al., 2014). We aimed to develop a computational approach to identify somatic events in non-coding regions that are both recurrent and functionally relevant in cancer in an unbiased, genome-wide scale. Somatic mutational density in cancer is known to correlate with various genetic and epigenetic features, including replication timing, gene expression levels and chromatin states (Lawrence et al., 2013; Schuster-Böckler & Lehner, 2012; Supek et al., 2014). We made use of this prior knowledge in our approach to correct for background mutational rates and to identify recurrently mutated regions in cancer genome. Our approach uses genomic windows and employs a list of covariates to select regions with comparable genomic background mutational rates. Among these, regions that are recurrently mutated in tumors are identified, functionally annotated and tested for potential mechanisms of oncogenesis (i.e. follow-up on target gene expression changes). We performed our initial analysis on a cohort of 698 cancer genomes (Alexandrov et al., 2013; unpublished). Our computational validations on a list of 209 unpublished medulloblastoma samples demonstrate that our findings are consistent with the previous knowledge on medulloblastoma biology. For example, we identified in an unbiased way recurrent mutations in: 1) WNT signaling pathway genes (CTNNB1, SMARCA4, PTCH1) as candidates in medulloblastoma of WNT subtype; 2) previously-described TERT promoter mutations. Additionally, we identified a promising novel candidate region on chromosome 11 in lung adenocarcinoma patients. We are currently developing and extending our pipeline to test for effects of parameters choice, including those recently published, on recurrent non-coding mutations identification: for example genomic window sizes (ranging between 50 bp and 10 mb), genetic and epigenetic features correlating with recurrent somatic mutations in cancer alone and in a combination. Citation Format: Vasilisa Rudneva, Simon Anders, Wolfgang Huber, Jan Korbel. A computational approach to identify recurrent somatic driver events in noncoding regions in human cancers. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-305. doi:10.1158/1538-7445.AM2015-LB-305