Sasha will be presenting a platform talk titled “The landscape of chromatin activity in renal cell carcinoma reveals thousands of germline regulatory variants with somatic interactions” at the American Society for Human Genetics 2017 meeting in Orlando. This talk focuses on new work, in collaboration with the Freedman lab, using allelic imbalance to understand the regulatory variants driving cancer risk. The talk is on Saturday, October 21 at 8:45am in Room 220B, Level 2. The full abstract of the talk is below.
Stick around for the exciting session on Transcriptome-wide Association Studies immediately after: Saturday, October 21 at 9:45-10:45am in Room 220B, Level 2.
ABSTRACT: The landscape of chromatin activity in renal cell carcinoma reveals thousands of germline regulatory variants with somatic interactions
Alexander Gusev, Matthew Freedman
The non-coding genome poses the next great challenge in understanding cancer development. In this work we investigated somatic and germline regulatory mechanisms in renal cell carcinoma (RCC) using H3k27ac ChIP-seq data in 10 matched tumor/normal samples and RNA-seq data from 496/66 tumor/normal samples. Unsupervised clustering of H3k27ac activity cleanly separated tumor from normal individuals, highlighting extensive epigenetic changes that occur after transformation. The H3k27ac signal was localized to 131,815 broad enhancer peaks, of which 101,397 were distal and did not overlap promoters. Consistent with their role in increasing transcription, tumor-specific enhancers overlapped genes with significantly higher tumor-specific expression; likewise, methylation probes (which are typically repressive) overlapping tumor-specific enhancers had significantly depleted tumor-specific activity. We identified 3,747 super-enhancers, of which 6 were observed in all tumors and none of the normal samples. These tumor-specific super-enhancers overlapped known and suspected RCC oncogenes such as EGLN3 and BHLHE41, implicating specific cis regulatory elements.
We developed a novel method to test each peak for allelic imbalance in binding (asbQTL) and evaluate tumor/normal differences in allelic imbalance (d-asbQTL) while accounting for structural variation and over-dispersion. These d-asbQTLs are a novel, functional approach to identify germline variants that interact with the somatic environment. At an FDR of 5%, we identified 1,356 unique asbQTL peaks in normal, 2,868 in tumor, and 1,054 d-asbQTLs (primarily imbalanced in tumor). This abundance of d-asbQTLs highlights the matched tumor/normal study design as a unique opportunity to identify putative cancer mechanisms that could not be observed in either study alone. d-asbQTL peaks were enriched for recurrent somatic mutations, indicative of positive selection and putative driver mechanisms. Intersecting significant H3k27ac asbQTLs with aseQTLs from tumor RNA-seq increased power, dramatically reduced the number of putative causal variants, and localized enhancers to upstream regions of target gene promoters. We are now integrating these regulatory variants with RCC GWAS data to identify specific cancer risk mechanisms.