September 13th, 2019
Topic: Integrating multi-omic data on the cloud yields insights into retrotransposon activity in cancer
Presenter: Wilson McKerrow, NYU School of Medicine, Manisha Ray, SBG
Abstract:
Comprehensively addressing many research questions require the use of data from multiple analytes (DNA, RNA, protein, imaging, etc). However, data is often collected, stored, and managed by different groups, and utilizing multiple large public datasets housed in different locations can be challenging. One area that is particularly in need of multi-modal data is that of retrotransposons, genomic elements that are capable of copying themselves to a new locus via an RNA intermediate. LINE-1 elements are the only retrotransposon that is autonomous and active in the human genome. LINE-1 elements contain an internal promotor, two open reading frames and a poly A 3’ UTR, and they are expressed during early development but become repressed through a combination of DNA and histone methylation as cells differentiate into somatic tissues. However, in many cancers, LINE-1 elements are dramatically de-repressed, possibly shaping tumor evolution. Fully understanding the role if LINE-1 elements in cancer requires analyzing genomic, transcriptomic, and proteomic data. In collaboration with Seven Bridges, we are re-analyzing CPTAC data to measure LINE-1 activity at RNA, protein, and insertion. To address the challenge of analyzing multiple data types, the Cancer Genomics Cloud, powered by Seven Bridges, developed a new way to access and compute on data held in two nodes of the Cancer Research Data Commons: the Genomic Data Commons and the Proteomic Data Commons. Our early focus has been on endometrial cancer as we observe robust LINE-1 expression in this cancer and have high quality data. We see good correlation between LINE-1 RNA and protein expression and between expression and the number of identified somatic LINE-1 insertions, indicating both that our quantifications are successful and that transcription is a key point of control for LINE -1 in cancer. Finally, after correcting for cancer purity, we see a correlation between LINE-1 expression interferon signatures, recapitulating in vitro results that LINE-1 expression leads to an induction of type 1 interferons.