Dr. Sadaf Alam (Swiss National Supercomputing Centre), Dr. Melissa Smith (Clemson University), Dr. Ross Walker (University of California, San Diego), Dr. Paul Crozier (Sandia National Laboratories), Duncan Poole (NVIDIA), and Carlos Sosa (Cray, Inc.). The future computing architectures, particularly the heterogeneous platforms, are posing new challenges for the scientific computing. We are working with several academic groups and hardware vendors for porting molecular dynamics simulations and molecular docking codes on graphical processing units (GPUs), many-core processors, low power ARM processors, Field-Programmable Gate Arrays (FPGAs). Our emphasis is on developing performance models that will allow optimal utilization of the heterogeneous computing resources for best scientific productivity. Codes including LAMMPS, AMBER and AutoDock are being optimized for heterogeneous architectures.
Dr. Chakra Chennubhotla (University of Pittsburgh). Jointly developing computational methodologies for identifying multi-scale motions and energy transfer in proteins. In particular, we have developed a novel method named Quasi-Anharmonic Analysis (QAA) based on the higher-order statistics of protein motions. Application of QAA has allowed the identification of functionally relevant protein motions and conformational sub-states. QAA has been successfully used for investigating several systems including ubiquitin, lysozyme, cyclophilin A, and adenylate kinase. Dr. Chakra's ongoing work is developing HOST4MD (Higher-Order Statistics Tooklit for Molecular Dynamics simulations), an automated package that allows identification of functionally relevant motions and conformational sub-states.
Dr. Sheldon Broedel (AthenaES, Baltimore). We are currently working with Dr. Broedel's company for creating the novel engineered enzymes. Based on identification of the enzyme energy networks for lipase B from Candida antarctica, mutant enzyme was created, purified, chemically modified and assayed. The goal of this work is to use novel protein engineering strategies for hyper-catalytic enzymes. We are continuing this collaboration for other enzymes.
Dr. Kenneth Herwig (Deputy Director, Neutron Scattering Sciences Division, ORNL) and Dr. Dean Myles (Director, Neutron Scattering Sciences Division, ORNL). This collaboration is joint computational-experimental investigation of protein structure, dynamics and function. Neutron scattering, X-ray crystallography and biophysical simulations are being used to investigate the protein motions at various time-scales. For the protein Rubredoxin, this strategy provided unique insights into the onset of function promoting motions. More recently, we are also using other strategies (including enzyme catalysis in organic solvents) to investigate the role of protein function promoting motions.
Dr. Elizabeth Howell (University of Tennessee, Knoxville). We are using joint experimental-computational strategy to understand the similarities and differences between the same enzyme reaction catalyzed by nonhomologous protein fold that do not share sequence or structural similarity. We are investigating the hydride transfer reaction catalyzed by the plasmid encoded dihydrofolate reductase (R67 DHFR). Based on our computational models of R67 dihydrofolate reductase enzyme, Dr. Howell's lab is designing mutation and experimentally verifying the role of protein and solvent motions in the enzyme kinetics. The comparison of the mechanism of R67 DHFR with the nonhomologous chromosomal DHFR (E. coli DHFR) is providing new insights into the role of structure and flexibility in enzyme catalysis.
Dr. Antonio Ferreira and Dr. Stephen White (St. Jude Children's Research Hospital). Developing high-throughput computational infrastructure for allosteric site recognition and drug screening. The energy flow in protein motions allows to identify and rank allosteric sites. We have developed a computational infrastructure that allows supercomputing machines for docking several hundred thousand compounds simultaneously against the identified allosteric sites.