MIE Seminar Presentation: Multiscale Models for Proteins: Biological Water as a Nanostructured Material by Dr. Jaydeep Bardhan
~~ The Department of Mechanical and Industrial Engineering Presents:
Dr. Jaydeep P. Bardhan
Department of Electrical and Computer Engineering, Northeastern University
Topic: Multiscale Models for Proteins: Biological Water as a Nanostructured Material
Date: Friday, April 18, 2014
Time: 10:00am to 11:30am (Refreshments from 9:30am to 10:00am)
Location: 342 Curry Student Center
Abstract: Protein structure and function depend strongly on the surrounding biological water (an aqueous electrolyte solvent). Molecular dynamics (MD) simulations naturally capture atomic-level details of protein-solvent interactions, but require such large computational resources as to be impractical for many studies. On the other hand, continuum solvent models (based on the macroscopic Poisson equation) are orders of magnitude faster than MD, but sometimes fail to predict experimental results because the underlying approximations are too strong--for instance, assuming that solvent molecules have zero size. In this talk, I will describe new multiscale mathematics, both models and numerical algorithms, that are transforming our ability to understand proteins in biological water; these efforts parallel important developments in modeling mechanics and structured materials, and the talk will highlight emerging opportunities for synergy. For example, to accurately model complicated protein-water interfaces (the atomistic--continuum interface), we use fast, parallel boundary-element method (BEM) simulations similar to those in linear elasticity and fracture mechanics. To model multiscale problems such as crowded protein solutions in pharmaceutical research and cell biology, we have designed novel operator approximations with tuneable accuracy, creating new possibilities to understand how manufacturing process variations affect device performance (i.e., uncertainty quantification). We are also adding crucial molecular realism to continuum models using rigorous statistical mechanics and atomistic MD simulations, advancing a new multiscale solvent modeling paradigm that is closely related to multiscale models in solid mechanics, e.g. gradient models and peridynamics. We have shown that the new models can help resolve a long-standing discrepancy between protein experiments and theory. Last, to address inverse problems in molecular design, I have developed PDE-constrained optimization techniques to speed up calculations in drug design and protein engineering by orders of magnitude. Combinations of these techniques offer new opportunities for simultaneous optimization of microscopic material structure and macroscopic properties.
Brief Bio: Dr. Bardhan is currently a Research Assistant Professor at Northeastern University in the Department of Electrical and Computer Engineering. He received his S.B., M.Eng., and Ph.D. degrees in Electrical Engineering from the Massachusetts Institute of Technology (MIT) in 2000, ’01, and ’06. As a doctoral student, he was the recipient of a Computational Science Graduate Fellowship award from the US Department of Energy (DOE). After his Ph.D., he spent three years at Argonne National Lab, the first two in the Mathematics and Computer Science Division as the Wilkinson Fellow in Scientific Computing, and one year in the Biosciences Division as an Argonne Scholar. During this period, he was named a Frederick A. Howes Scholar, in honor of the late DOE Applied Math program manager. Prior to joining Northeastern, he was an Assistant Professor of Molecular Biophysics and Physiology at Rush University Medical Center in Chicago, where he also held a visiting appointment in the Bioengineering Department at the University of Illinois, Chicago.
Friday, April 18, 2014 at 10:00am to 11:30am
Curry Student Center, 342
346 Huntington Avenue, Boston, MA, Boston