A major research focus for gaining insight into the molecular basis for cancer, with the goal of identifying new treatments, is the systematic profiling of cancer cells. It has resulted in large volumes of data that include the compound potency and gene expression data. To more effectively find treatments, the data must be integrated and associated in context sensitive ways. The overall objective of this project is to create a web-based, high-performance visual analytics platform for the exploration of compound activity and molecular profile data, through interactive visualization that supports human analytical reasoning. The platform will demonstrate how visual exploration of cancer cell-based data that include genomics, proteomics, cytotoxic potency, and other research data can be used to identify candidate targets and drugs for cancer chemotherapy, and determine factors that contribute to chemoresistance and sensitivity.
The specific aims are the following: 1) Improve the ability to navigate, search, and directly explore large data in highly-coordinated multi-view visualizations. The heatmap will incorporate interaction mechanisms from zooming interfaces and animation. A new visualization for results of the Tau Path statistical method will explore pairs of variables, such as compound-gene, where significant associations have been found in some, but not all of the cell lines. 2) Develop a high-performance analytics server for multicore systems. The server will execute concurrent components for significance tests, clustering, correlation, regression, and association analysis. 3) Extend the new Tau Path statistical method. The extensions include improved identification of the subpopulation, conditional testing to identify subpopulations under which associations are significantly greater than for the whole population, and extension to larger populations, such as all compounds of interest, so as to prioritize drug candidates 4) Integrate resources with the platform to enable findings to be rapidly validated and provide additional insights. The resources include biological pathways systems (e.g. KEGG), the Gene Ontology annotations (GO), biological screening databases (e.g. PubChem), and chemoinformatics services (e.g. OpenTox). The impact of the resulting visual analytics platform will be to more rapidly identify hypotheses that lead to the discovery of new cancer treatments. These hypotheses include compounds that may serve as leads in a drug discovery program seeking new chemotherapy and genes that modulate drug potencies in selected cell lines. The project will lead to a visual analytics platform for cancer research and curated public content.
Cancer is a major cause of mortality throughout the world. Many types of cancers lack effective treatments, and new treatment approaches have the potential to positively impact large patient populations. This project will build a new high-performance visual analytics platform to aid in identifying biomarkers predictive of drug efficacy for optimizing cancer chemotherapy.