Model-Based Drug Design: Lessons Learned

Model-Based Drug Design: Lessons Learned

Thursday, January 11, 2007

The New York Academy of Sciences

Presented By


Organizer: Gustavo Stolovitzky, IBM

The Systems Biology Discussion Group aims to explore the complex and often dynamic interdependencies between gene regulation, cellular signaling, and cellular metabolism. Meetings of the group focus on ongoing efforts to formulate complex functional queries about the cell, in a genomic context, to explore them in silico, and to produce testable hypothesis about the physiological impact of individual gene manipulations.




David de Graaf, Pfizer. "Systems Biology at Pfizer: From Physiology to Pharmacology for p38 inhibitors."

An orally available inhibitor of p38 is one of the holy grails of today's biotech and pharma, for its expected efficacy in treating Rheumatoid arthritis. Approximately 20 compounds directed against p38 have been brought to the clinic by a variety of companies. Common adverse events are reported with small increases in specific liver enzymes and an acne-like rash being the most common. We are using Systems Biology approaches to identify mechanistic links between the molecular level and these particular adverse events.
In today's talk I will outline our approach using both mechanistic representations for the simulation of pathway behaviors, as well as large statistical models to provide a link between complex phenotypic behaviors and the molecular events underlying them. This type of program can be used to support a drug discovery and development program from inception and target choice to patient stratification and biomarker selection.

Birgit Schoeberl, Ph.D., Director, Network Biology, Merrimack Pharmaceuticals. "Systems Analysis of Cellular Responses to Therapeutic Interventions"

A crucial challenge is the development of approaches toward prediction of how molecular interventions might affect cell phenotypic function, across multiple cell types and heterogeneous patient populations. This presentation describes nascent attempts to construct predictive computational models for cell behavior from quantitative protein-level experimental studies, and thoughts about their application in the clinic.

Iya Khalil, Gene Network Sciences.
"Reverse Engineering and Forward Simulations of Regulatory Networks in Preclinical and Clinical Applications"

A major challenge in drug development is discovering the biological mechanisms of how compounds work, particularly in the context of specific genetic backgrounds. A new approach for determining compound mechanism of action is to reverse engineer the mechanism of action based upon cumulative observations of compound affects on cellular components in high-throughput experiments. Here we present a computational data-driven approach for learning causal models from molecular profiling and phenotypic data, and then systematically and rapidly interrogating these models via high-throughput forward simulations. We apply our methods to determine causal drivers of efficacy underlying multi-kinase inhibitors. We then show that this approach can correctly predict causal links to key players in ErbB signaling, recapitulate known differences in compound mechanism of action in different cancer cell types, and provide new insights that differentiate between mechanisms of action of closely related ErbB kinase inhibitors. We then discuss the extension of this approach to clinical applications.