A Pan-Cancer Network Analysis with Integration of miRNA-Gene Targeting for Multiomics Datasets
Volume 19, Issue 4 (2021), pp. 555–568
Pub. online: 16 August 2021
Type: Statistical Data Science
Received
12 March 2021
12 March 2021
Accepted
17 July 2021
17 July 2021
Published
16 August 2021
16 August 2021
Abstract
Large-scale genomics studies provide researchers with access to extensive datasets with extensive detail and unprecedented scope that encompasses not only genes, but also more experimental functional units, including non-coding microRNAs (miRNAs). In order to analyze these high-fidelity data while remaining faithful to the underlying biology, statistical methods are necessary that can reflect the full range of understanding in contemporary molecular biology, while remaining flexible enough to analyze a wide range of data and complex phenomena. Leveraging multiple omics datasets, miRNA-gene targets as well as signaling pathway topology, we present an integrative linear model to analyze signaling pathways. Specifically, we use a mixed linear model to characterize tumor and healthy tissue, and execute statistical significance testing to identify pathway disturbances. In this paper, pan-cancer analysis is performed for a wide range of signaling pathways. We discuss specific findings from this analysis, as well as an interactive data visualization available for public consumption that contains the full range of our analytic findings.
Supplementary material
Supplementary MaterialSupplementary Materials include descriptions for data and software.