Professor Ron Elber

  Cornell University

Department of Computer Science

4130 Upson Hall, Ithaca,  NY 14853

Email: ron@cs.cornell.edu

Phone: (607) 255-7416

 FAX: (607) 255-4428

General Field of Investigation:

Structure and dynamics of biological molecules: Theory, algorithm developments and computer simulations. Software for molecular dynamics simulation -- MOIL and for protein recognition -- LOOPP

Figure: The gramicidin ion channel (sticks and green ribbon) and a "permion"  transported through it. A permion is a collective "quasi-particle" that consists of about 8 water molecules (red balls with white spots attached) and a sodium ion (the small pink ball in the middle) that moves coherently   (like in a queue) through the channel. A reaction path algorithm was   implemented into the Molecular Dynamics package MOIL [9] and was used to study the properties of the permion. The picture was made with the program MOIL-View written by Carlos Simmerling [10].

Topics of Research:

Ron Elber’s research focuses on algorithm development and computer simulations of the structure and dynamics of biomolecules.

Currently active research directions include

• The design and application of protein folding potentials. Linear programming techniques are used to find optimal sets of parameters that recognize the native folds of proteins. Potentials are designed for maximum accuracy and optimal computational efficiency. We are studying a hierarchy of potentials, some functional forms are trained to the highest accuracy and other formulations are less accurate but are trained to search more rapidly for plausible conformations.
• The design and application of algorithms to compute long time dynamics of biological molecules. Algorithms that are based on optimization of functionals are investigated as an alternative to the usual approach of solving initial value differential equations. Approximate stable trajectories of very broad time scales can be obtained. Time steps larger by a factor of million compared to the common simulation algorithms were used successfully.
• Mean-field global-optimization as a tool to determine peptide structures in solution. We developed and applied mean field simulation techniques that make it possible to compute peptide conformations in explicit solvent environment. Structures of peptides of lengths smaller than ten amino acids were determined and compared favorably with experiment.
• Searching for plausible diffusion pathways from buried active sites. In a number of proteins (e.g. the globins) the active site is buried in the protein interior and is not accessible to a solvated ligand. We developed efficient search strategies for plausible diffusion pathways and gateways, which are induced by thermal fluctuations of the protein. The search method, the Locally Enhanced Sampling approach (LES), enables rapid exploration of small fraction of space which is of prime interest (e.g., the accessible volume at the active site).
• Reaction path studies of slow processes in proteins. We introduced novel computational approaches that make it possible to calculate minimum energy paths for complex processes with motions of thousands of atoms, like the R to T transition in hemoglobin, or ion permeation through a membrane channel. Statistical approaches, which are based on the availability of the reaction paths, are finally used to estimate the time scales measured experimentally.