Software tools for visual analysis of cancer genomics data in the context of pathways

Abstract

Information visualization is concerned with effective visual presentation of abstract information, which reinforces human cognition. Graphs are structures that are well suited to represent relational information. Graph visualization is vital since the underlying relational information of the graph provides fine analysis and comprehension opportunities. Biological pathway visualization is one of the most popular areas, where graph visualization is highly favored. Interactive analysis and visualization of cancer related pathways in the context of genomic data, such as those available through the TCGA project, might reveal valuable information for scientists about disease conditions and potential causes. As the size and complexity of such cancer pathways and associated genomic data increase, exchangeable in-silico representations and their effective, enhanced visualizations, and complexity management become crucial for effective analysis of such data to potentially discover cause-effect relations. In this thesis, we designed and implemented software solutions to visualize cancer genomics data in the context of networks from simple gene interaction networks to process description diagrams within the cBioPortal for Cancer Genomics (cBioPortal). cBioPortal is a popular web portal, getting about 60.000 visits globally per month, providing visualization, analysis and download of largescale cancer genomics data sets. The network view in cBioPortal presents neighborhood of genes of interest. The alteration data is overlaid on the network with numerous ways to filter and manage complexity of the network (e.g. by alteration percentage or by type or source of the interactions). Upon demand, the user can obtain a more detailed, mechanistic view of the interactions among gene pairs, from Pathway Commons database with a live query using the SBGN process description notation. Finally, we also developed a new pathway visualization component, specifically for cancer pathways, using a uniform notation found in TCGA cancer publications. This tool also facilitates curation of pathways from scratch with support for collaborative editing.