Back to Pre-Meeting Webinars

Presented by: John Bachman, Fellow in Therapeutic Science Laboratory of Systems Pharmacology, Harvard Medical School

Title of Presentation

"Machine-assisted modeling to accelerate biomedical discovery"

This Webinar is posted on the MSM YouTube Channel

Abstract

Biology is currently grappling with the challenge of integrating large datasets with a body of scientific knowledge that has grown too large and complex for any single scientist to read or understand. The size and scope of large-scale data compendia (“big data”) has created analytical challenges that require new solutions, ideally ones that make use of aggregated scientific knowledge. To address this problem we have developed the Integrated Network and Dynamical Reasoning Assembler (INDRA), a system that automatically assembles mechanistic models from pathway databases, literature, and expert knowledge expressed in natural language. INDRA draws on six existing natural language processing systems and uses a modular architecture to build different types of models from a variety of sources. This framework has a number of applications in the biomedical modeling and discovery process. Domain experts can express biological knowledge directly in natural language and build executable models automatically, bypassing the step of manually encoding assumptions in a formal modeling language. At a larger scale, INDRA can construct causal graphs from literature-scale text mining to identify candidate explanations of observations in large experimental datasets. This process can be automated, with the system automatically scanning the literature daily for new mechanistic findings, separating new from known information, and checking whether the new findings impact the ability of the model to explain existing observations. Finally, we have embedded capabilities for natural language modeling, simulation, and causal network search in a human-machine dialogue system, allowing scientists to ask questions, and receive answers, about biological mechanisms in English. We are currently applying these tools to discovery efforts in cancer pharmacology and neurodegenerative disease.

Biosketch

Dr. John Bachman is a Fellow in Therapeutic Science at Harvard Medical School's Laboratory of Systems Pharmacology. His research focuses on the development of computational tools for understanding the behavior of complex biological systems, and the application of these tools to studying problems of cellular decision-making in health and disease. In his most recent work he co-developed the Integrated Network and Dynamical Reasoning Assembler (INDRA) to automate the construction of explanatory biological models from natural language and scientifc literature. John received his Ph.D. in Systems Biology from Harvard University working in the lab of Dr. Peter Sorger, where he combined wet-lab experimentation and computational modeling to address unresolved mechanistic questions in programmed cell death. Before obtaining his Ph.D. John worked as a scientist for four years at the Cambridge, MA, research software company Charles River Analytics. At CRA he worked on several projects for the Army and Air Force Research Labs using simulation and knowledge management tools to improve human decision making.

This webinar is part of a series of recordings for the ML-MSM Meeting, see complete list of Pre-Meeting Webinars