Stochastic methods for modeling molecular-protein interactions are entirely new approaches to the important biological goal of simulating cellular biology in silico. Though great progress has been made in this direction by computational biologists over the past 15 years, the goal of siliconizing cellular molecular interactions still remains remote. More precisely, the current level of model realism has led to a plateau in the prediction accuracy of molecular interactions. This motivates the development of novel dynamic models that incorporate more extensive biological details and can add layers of realism to the simulations. However, such models are computationally daunting, so that it is critical to develop efficient and accurate computational methods. The purpose is to augment molecular-level understanding and simulation of biological interactions. In particular, by exploiting novel developments in stochastic optimi?ation, the investigators shall significantly improve the prediction of interactions by adding a new dimension of realism. However, for many practical cases the stochastic objective function will become a high dimensional, nonGaussian, nonlinear random field that will be computationally very challenging to optimize. This is a hard problem that the investigators plan to address by developing novel Uncertainty Quantification (UQ) mathematical theory.
The specific aims are to: (i) Develop a compact dynamic surrogate model of the stochastic objective function that incorporates the molecular structure uncertainty and molecular properties such as the electrostatic fields by solving the nonlinear Poisson Boltzmann (PB) equation. The stochastic optimization is solved efficiently with a surrogate model. (ii) Analyze the complex analytic regularity properties of the solution of the nonlinear Poisson-Boltzmann equation (and the other molecular properties) with respect to the probabilistic molecular conformation model. (iii) Develop convergence rates of the surrogate model from the complex analytic regularity with respect to the number of realizations of the protein structure (computational complexity). Most protein interactions models based on molecular. structure assume a rigid shape thus leading to erroneous predictions. The investigators propose to significantly improve the prediction of protein interactions by incorporating dynamic uncertainty of the molecular conformational shape. The theory and application of UQ to protein interactions is at its infancy.

Public Health Relevance

The investigators feel that this approach will change the level of ability of computers to simulate cellular processes by exploiting the theory of modern uncertainty quantification. This work will have a direct impact in drug design and potential synthesis of antibodies.

Agency
National Institute of Health (NIH)
Institute
National Institute of General Medical Sciences (NIGMS)
Type
Research Project (R01)
Project #
1R01GM131409-01
Application #
9674591
Study Section
Special Emphasis Panel (ZGM1)
Program Officer
Lyster, Peter
Project Start
2018-08-01
Project End
2021-04-30
Budget Start
2018-08-01
Budget End
2019-04-30
Support Year
1
Fiscal Year
2018
Total Cost
Indirect Cost
Name
Boston University
Department
Biostatistics & Other Math Sci
Type
Schools of Arts and Sciences
DUNS #
049435266
City
Boston
State
MA
Country
United States
Zip Code