Open Collections

Network-based integrative analysis of multi-omic data

University of British Columbia

The rise of high-throughput biology has brought an increase in generation of large datasets such as genomics, transcriptomics, proteomics, and metabolomics: “omics” data. While many biological studies now assay multiple omics types to assess biological function, the analysis of these datasets is typically undertaken separately, contrary to our understanding of how biological systems function. While efforts have been undertaken to integrate these data types, intuitive methodologies that take advantage of modern curated biological databases are lacking. Here I present a methodology for network-based integrative analysis of multi-omic data. This method leverages the power of curated interactome databases and biological network analysis to produce multi-omic biological interaction networks for integrative analysis. The integration of metabolomics data with transcriptomics and proteomics data was enabled by identifying metabolite-protein interactions using MetaBridge, a novel tool that I developed, described here. Identification of these metabolite-protein interactions was shown to facilitate the leveraging of powerful curated protein-protein interaction (PPI) databases such as InnateDB to generate metabolome-centric PPI networks. Such PPI networks accurately encapsulate biological function and enable downstream analysis and dimensionality reduction using proven network analysis techniques. These metabolomics-derived PPI networks could then be integrated with proteomics and transcriptomics data to create multi-omic networks, which provided insights into biological function and could be mined for novel biological insights that would not otherwise be captured by any single omics type. I demonstrated two applications of this methodology to multi-omic datasets. First, I showed how separate gene expression and metabolite signatures for predicting sepsis could be integrated to reveal novel targets for study, demonstrating the utility of this method for hypothesis generation. Second, I demonstrated tri-omic integration of metabolomics, proteomics, and transcriptomics data from neonates in the first week of life. This revealed that network-based multi-omic integration provided consensus on commonly dysregulated biological functions and facilitated novel insights into biological changes.

Text

2018-10-15

Attribution 4.0 International

http://hdl.handle.net/2429/67564

Bioinformatics

University of British Columbia

UBCV

http://creativecommons.org/licenses/by/4.0/