We use computational approaches to study structures and functions of biomolecules. Our final goal is to interpret the structural and functional information encoded in genomes and to understand life at the most fundamental level based on physical and chemical principles. Our current research focuses on developing reliable and efficient methods to study biomolecular structures, functions, and intermolecular interactions at atomic detail, and applying our new methods to understand and predict the relations between the sequences, structures and functions of these molecules.

Protein Folding Mechanism and Structure Prediction

The ability of proteins to perform various biological functions is attributed to their unique three-dimensional structures which are predetermined by their amino acid sequences. The amazing feat that these molecules have to accomplish to assemble themselves quickly and reliably has puzzled scientists in many different fields for decades. The so-called protein folding problem involves the fundamental issue to understand the folding mechanism and the challenging final goal to predict three-dimensional structures from amino acid sequences. It is often regarded as the second half of genomics.

Our research in this area involves understanding protein folding mechanism in both thermodynamic and kinetic details by atomistic molecular dynamics simulations (see below). The grand challenge is how to simulate atomic-detailed phenomena which occurs on time scales of microseconds to milliseconds. The second aspect of our research involves developing methods to predict protein structures accurately, efficiently and automatically to meet the challenge of overwhelming genomic data.

Protein-Protein and Protein-Ligand Interactions

Structure determination is only the first half of biochemistry. Function annotation is the other more important half. This requires that we study how biomolecules interact with each other and how such interactions assist functioning. The knowledge is often helpful, sometimes critical, in developing new medicines. Computational studies on intermolecular interactions can provide many insights to the understanding. Computational findings are also much easier to be converted into new virtual compounds which may eventually lead to new drugs.

Our goals in this area are to reduce the gap between the high-end physical chemical studies of the biomolecule associations and the low-end rigid-body docking studies in virtual screening of compound libraries, and to be able to predict protein-protein and protein-ligand interactions based on structural and sequence information of proteins. A series of reliable and efficient tools will be needed to reach these goals. These efforts will be complimented by applications to interesting biochemical systems.

Simulation Methodologies for Computational Biology and Chemistry

A common problem in the computational biochemistry is that atomic level models for liquid-state processes can be simulated for only a few tens of nanoseconds with the current computing power, while biochemical processes, such as protein folding and binding are often on the time scale of microseconds or longer. Much simpler models based on lattice or off-lattice representations with residue-level resolution have been developed to study such processes. Quite often, an atomic-detailed description of biomolecules is important in elucidating their structures and functions. Therefore a balance must be achieved in developing models with simple interactions that permit fast calculations without the loss of the important atomic detail of biomolecules. A continuum solvent model based on the classical electrostatic theory has been developed for this purpose. The model, termed Poisson-Boltzmann continuum solvent, has been widely accepted as a reliable way to treat solution energetics and processes of biomolecules. However, the numerical solution of the Poisson-Boltzmann (PB) equation has been a bottleneck in limiting their applications with static structures only.

We have dramatically improved the efficiency of a finite-difference (FD) numerical solution of the PB equation. It is now possible to use FDPB in molecular dynamics simulations with efficiency comparable to vacuum simulations. We have adopted the strategy of Particle-Particle and Particle-Mesh to achieve the desired efficiency and accuracy in computing the total electrostatic energy and forces in a molecular mechanics simulation. The molecular dynamics (MD) simulation with the current FDPB method has been demonstrated to be robust and stable for many systems studied. We are working to improve PBMD to make it more accurate and more efficient and port it to parallel platforms.

Research Opportunities

Graduate research can be arranged through the following programs: Structural Biology and Molecular Biophysics, Mathematical, Computational and Systems Biology, and Biomedical Engineering at University of California, Irvine.

Postdoctoral positions are available in my group from time to time. Research involves development and application of new tools for computational studies of biomolecules. Candidates with a background in biophysics, chemistry, physics, material science, and biomedical/chemical engineering are encouraged to apply. Previous experience in computational research is required. Experience in scientific programming and molecular dynamics simulations is a plus. Please send a CV, a statement of interests, and two letters of reference to:

Dr. Ray Luo
Department of Molecular Biology and Biochemistry
Department of Information and Computer Science
Center for Biomedical Engineering
University of California
Irvine, CA 92697-3900, USA
rluo@uci.edu

Please send comments to: rluo@uci.edu