News & Events / BME Seminar Series / Past Seminar Speakers / 2018-19Hasan Otu
BME Seminar Series Winter 2019
Thursday, January 17, 2019 at 4-5 pm
Tech L361
Host: Professor Jonathan Rivnay
Professor of Electrical and Computer Engineering, University of Nebraska-Lincoln
Probabalistic Graph Models for Biological Data Analysis Uisng External Knowledge
Networks involving all direct and indirect interactions between genes and/or gene products (the interactome) can be used to understand biological pathways and disease mechanisms. From a data analysis perspective, the problem evolves in two directions: (i) given high throughput biological data (HTBD), which known pathways best explain the observed data, (ii) given HTBD, what are the interaction networks that can be deduced? The first question has traditionally been answered using enrichment-type approaches to find pathways that are “active” based on HTBD. The second question has typically seen attempts that use linear measures, which often do not utilize the existing knowledge of interaction between macromolecules. We have developed methods that use probabilistic graph representations to answer these two questions. For the first part, we model the topology of the pathway to identify its ability to explain the given HTBD. For the second part, we use external biological knowledge that supplements the inferences coming from the observed HTBD to discover interaction networks. In this talk, I will describe the two approaches with results based on synthetic, simulated, and real HTBD.