Research

De-novo Protein Design

Understanding Protein Catalytic Activity through Dynamics and Thermodynamics

De-Novo enzyme design involves designing enzymes from scratch in order to perform certain specific activities. However most of these enzymes are not as efficient as their naturally occurring counterparts. The state of the art approaches in this field are mostly structure based and view the protein active site as static in nature. Consequently, these approaches do not take into account the dynamics of the residue motions and thermodynamic effects manifested by conformational changes in the enzyme.

Our lab investigates these effects by using Molecular Dynamics and Monte Carlo simulations to correlate dynamic and thermodynamic metrics with enzyme catalytic activity.

Researchers: Saurabh, Asmit

Intrinsically Disordered Proteins

Intrinsically disordered proteins (IDPs) are a subset of proteinsthat lack a dominant in vitro conformation of their secondary and tertiary structures. IDPs are involved in cell signaling and have been linked to a variety of neurodegenerative diseases.
A thorough understanding of their dynamics is thus key to developing effective treatments and studying the associated cellular processes. Their propensity to sample a very large conformational space, however, greatly increases the computational complexity of any de novo molecular dynamics simulations. In the Head-Gordon lab we are working on developing new sampling algorithms that, in conjunction with modern parallel computing technologies and existing NMR spectroscopic data, will reduce this simuation expense and will eventually lead to efficient computational beamlines for the study of IDP structural biology.

Modeling Electrostatic Phenomena

Electrostatic interactions play an important role in molecular simulations. The Poisson-Boltzmann equation (PBE), which describes the charge density using mean-field approximations, can give very accurate results for the electrostatic potential of various systems. Our group has developed PB-SAM, a semi-analytical method for solving a linearized PBE based on ideas from the fast multipole method (FMM) and boundary element (BE) method. Compared with other PBE solvers such as finite-difference (FD) and finite-element (FE), PB-SAM can handle molecules with arbitrary shapes, realize better accuracy, has flexible memory management and a significantly reduced computational cost.

We are currently using PB-SAM combined with Brownian dynamics (BD) simulations to study proton transport through a Nafion® membrane and electro-osmotic drag in polymer electrolyte membrane (PEM) fuel cells.

Researchers: Lisa, David Brooks

 

Force Field Development

The energy of a system of molecules is non-additive. That is, the interaction energy of a molecule with surrounding molecules depends on more than just pairwise interaction with surrounding molecules. For example, in electrostatics the interaction of fixed electrostatics is a pairwise interaction while the induced dipoles on the atoms are a many-body effect. The energy of a system can be evaluated by a many-body expansion of the interaction energies to give the full energy of the system as shown:

                                                    Etot = E(1) + E(2) + E(3) + … + E(N)

In quantum chemistry simulations, it has been known since the early 1970’s that this energy is dominated by the 1, 2, and 3 body energy terms. The many- body energy expansion can be truncated to the three-body energy term and give nearly the same energy as the full evaluation of the many-body expansion (i.e. Etot ≈ E(1) + E(2) + E(3)). This effect has recently been exploited in quantum simulations to reduce the time for the computation of the full energy of a system by distributing a large number of simple jobs in parallel*. Similarly, we aim to incorporate this approximation in a classical polarizable model (AMOEBA) for the evaluation of the many-body energy term which arises from polarization. The ultimate benefit of this research is that polarizable models could be evaluated faster in a trivially parallelizable way.

Additionally, we are currently investigating thermalized Drude oscillators or the “charge on a spring” model to efficiently introduce polarization into molecular simulation.  The thermalized Drude oscillator is a small charge attached by a harmonic oscillator to an atomic center that can react to and influence its environment, mimicking polarization.  By coupling the oscillations to a cold thermostat we hope to achieve near self-consistent field (SCF) dynamics without computationally expensive SCF iterations.

Researchers: Omar, Alex

* U. Gora, W. Cencek, R. Podeszwa, and K. Szalewicz, J. Chem. Phys. 135

Disease Aggregation

Alzeihmer’s disease, a common neuro-degenerative disease, is characterized by the presence of soluble Amyloid- beta peptide in the brain. Amyloid-beta, being a intrinsically disordered peptide, can adopt many tertiary conformations in the brain which cannot be captured by experiments alone. Our lab combines the experimentally observable NMR data of amyloid monomer and oligomers with de novo molecular dynamic simulations to predict the ensemble of conformations that amyloid-beta can adopt in the brain. Our lab has also developed a coarse-grained protein model and used it to determine conformations of the amyloid fibrils. We are currently investigating possible amyloid-beta oligomer structures.

Researchers: Sukanya

J. Chem. Theory. Comput. 2010, 6, 2214-2224.
J. Phys. Chem. B. 2005, 109, 24244-24253

Intrinsically Disordered Proteins

Intrinsically disordered proteins (IDPs) are a subset of proteins that lack a dominant in vitro conformation of their secondary and tertiary structures. IDPs are involved in cell signaling and have been linked to a variety of neurodegenerative diseases. A thorough understanding of their dynamics is thus key to developing effective treatments and studying the associated cellular processes. Their propensity to sample a very large conformational space, however, greatly increases the computational complexity of any de novo molecular dynamics simulations. In the Head-Gordon lab we are working on developing new sampling algorithms that, in conjunction with modern parallel computing technologies and existing NMR spectroscopic data, will reduce this simuation expense and will eventually lead to efficient computational beamlines for the study of IDP structural biology.

Supercritical Water

Supercritical water is a medium for chemical reactions such as oxidation of hazardous wastes, novel materials synthesis, synthetic fuels production, and biomass processing. However, a class of water with similar phase behavior may represent a greener alternative: confined water, or water residing in spaces of nanoscopic dimensions next to hydrophobic surfaces. Confined water, like supercritical water, exhibits decreased dielectric constants and large changes in density in response to relatively small changes in pressure. These observations suggest that the phase behavior of confined water can be mapped onto the supercritical region of bulk water. Advanced, inexpensive computational methods are necessary to determine how varying wall hydrophobicity and confinement geometry influences the supercritical-like behavior. Non- polarizable models are sufficient for some condensed-matter phenomena but fall short in problems involving solvation and phase transitions. Therefore, models which accurately describe confined water must incorporate polarizability.