Abstract 3038: Bridging the gap between NGS data and its usability: cancer gene discovery through massive-scale transcriptomic analyses and development of a powerful web-tool for dissemination of these findings

Abstract

Abstract Background: In a prior transcriptomic analysis focused on lncRNA interrogation using some of these data, we discovered tens of thousands of novel lncRNAs (Iyer and Niknafs et al, Nature Genetics, 2015). Building on this analysis, we recently developed a markedly improved bioinformatics tool for novel gene/isoform discovery from massive RNA-seq datasets named TACO (Niknafs et al, Nature Methods, In press, tacorna.github.io). TACO produces high-fidelity transcript structure predictions from large RNA-seq datasets. We now set out to comprehensively leverage both TACO and the publicly available RNA-seq data for discovery of novel transcriptional cancer associations. Additionally, in order to widely disseminate these findings in an accessible manner, we have built a web-tool that provides the scientific community access to these data and analyses. Methods: We have downloaded, curated, and processed 23,623 RNA-seq samples largely from the TCGA, ICGC, GTEx, and CCLE, comprising 37 tissue types and over 35 cancer types. RNA-seq data processing was performed using STAR, Cufflinks, Kallisto, and TACO. The web tool for visualization and access to these data and analyses was built using a JavaScript-based server infrastructure (Noje.js) and a relational PostgreSQL database. Results: Generation of a consensus transcriptome from this large-scale RNA-seq dataset via TACO resulted in the discovery of tens of thousands of novel transcriptional elements, including intergenic non-coding RNAs and novel splice isoforms of known genes. Such an expansive RNA-seq cohort that includes many normal tissue samples enabled statistically powerful cancer association expression analyses that revealed a myriad of novel cancer genes, especially in tissues for which there was previously little-to-no normal tissue RNA-seq data (e.g., brain and pancreas). Many of these novel transcriptional elements discovered using TACO were also found to be cancer associated. We have built a web-tool to facilitate further analysis and discovery using these data and analyses by the scientific community. The web-tool provides a powerful and intuitive interface for researchers with little-to-no bioinformatics expertise to leverage large-scale RNA-seq datasets. Conclusion: Here we present the largest reported compendium of RNA-seq data, and reveal many novel cancer gene associations. Using a new, powerful gene discovery tool, TACO, we identify a multitude of novel transcriptional elements that are also cancer associated. Despite the abundance of publicly available RNA-seq data, necessary computing resources, data storage, and bioinformatics expertise are barriers to usability of these data by scientific community. Our RNA-seq expression web-tool bridges this gap, and enables users to powerfully interrogate cancer expression across dozens of tumor and tissue types. Citation Format: Yashar Niknafs, Nicholas Molen, Balaji Pandian, Matthew Iyer, Arul Chinnaiyan. Bridging the gap between NGS data and its usability: cancer gene discovery through massive-scale transcriptomic analyses and development of a powerful web-tool for dissemination of these findings [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 3038. doi:10.1158/1538-7445.AM2017-3038