OREGON STATE UNIVERSITY

You are here

An integrative computational approach to understanding cell type-specific gene regulatory networks, and its application to cardiovascular disease

KEC 1001
2013-12-03 00:00:00
Speaker Information
Stephen Ramsey
Assistant Professor
Department of Biomedical Sciences / School of EECS
Oregon State University

Within cells, gene expression is controlled by networks consisting of sequence elements in the genome and combinations of specialized regulatory molecules, called transcription factors, that bind these sequence elements. Computationally mapping these networks is difficult because of their scale, interconnectedness, and dependence on the cellular context. I am using an integrative computational approach to study gene regulatory networks in mammals, that incorporates both gene expression profiling and cell type-specific, genome-wide measurements of regulatory sequence elements. This approach, which I call "Regulatory Element Machine-Learning to Infer Networks Instructing Specific Cellular Expression (REMINISCE)", combines a machine-learning algorithm for pinpointing regulatory regions in the genome with a Bayesian statistical approach to identify transcription factors that may be associated with a given cellular response. I tested the approach on data from macrophages (which are a key cell type of the innate immune system and play an important role in cardiovascular disease) that were treated with an oxidized type of "bad cholesterol" (oxidized low-density lipoprotein) that is thought to promote artery plaque formation. REMINISCE had greater statistical power for identifying the relevant transcription factors in the macrophage gene expression responses, than the standard computational approach. We tested the function of one of these transcription factors, Activating Transcription Factor 3 (ATF3), in the context of a mouse model of arterial plaque build-up, and found that disruption of the gene for ATF3 increased the amount of plaque. These findings illustrate the benefit of gene regulation-focused, integrative computational approaches to uncover regulators in disease-associated gene networks. My talk will not assume any biology background and aims to convey the exciting potential for computational biology to revolutionize biomedicine.

Speaker Bio

Stephen Ramsey is a computational systems biologist who recently joined the Department of Biomedical Sciences (primary appointment) and the School of Electrical Engineering and Computer Science (secondary appointment) at OSU.  Out of his graduate training in physics, Stephen developed an interest in scientific computing and its applications in the study of complex systems. Following a postdoctoral appointment where he worked on computational methods for genome mapping in the Human Genome Project, Stephen joined the Institute for Systems Biology, where he developed integrative computational approaches to map gene regulatory networks. Since 2010, Stephen has worked on an NIH-funded project to apply these methods to study atherosclerosis. More broadly, Stephen’s research program aims to develop and apply data-driven, integrative computational approaches to gain insights into the immune system-specific regulatory networks that control cellular behavior in health and in disease. On the computational side, Dr. Ramsey is interested in applications of machine learning, Bayesian statistical modeling, and dynamical modeling in biology.