Research Nuggets

On the electrostatic component of protein-protein binding free energy

Emil Alexov, Professor of Physics at Clemson University, and his group present a comprehensive statistical analysis of the sensitivity of the electrostatic component of binding free energy (ΔΔGel) with respect with different force fields (Charmm, Amber, and OPLS), different values of the internal dielectric constant, and different presentations of molecular surface (different values of the probe radius). The minimization of the structures used in this work are made possible by the Condor pool deployed and maintained by Clemson Computing and Information Technology. See the full paper.

Estimation and Inference about Efficiency in Production Settings

Paul Wilson, Professor of Economics at Clemson University, uses the Palmetto cluster to undertake data- and computationally-intensive methods for nonparametric estimation that would have otherwise been impossible. The new resources are helpful for theoretical work aimed at developing new methods for estimation and inference about efficiency in production settings. Dr. Wilson is now able to apply these new methods while making more efficient use the very large amounts of data available on U.S. commercial banks to examine industry structure.

Protein Interaction with Synthetic Materials

Rober Latour, Professor and Endowed Chair in Bioengineering, Clemson University, uses the Palmetto cluster to conduct research to investigate how proteins in the body interact with the surfaces of synthetic materials. The long-term goal of this research is design of medical implants with improved biocompatibility.

Dr. Latour’s research group performs molecular dynamics simulations on proteins and polymers using the CHARMM program under 3 separate NIH grants. These simulations require a large number of CPUs for parallel processing runs using replical-exchange molecular dynamics (REMD) with a program called MMTSB for REMD control.

Effects of School Characteristics and Parents' Work Behaviors on Children's Performance in School

Thomas Mroz, Professor of Economics at Clemson University, uses the Palmetto "condo" computer cluster in his research on the effects of school characteristics and parents' work behaviors on children's performance in school and their intellectual development. A parent's choice of a school district to live in defines both a set of characteristics for the schools available to their children and a set of possible labor market opportunities for the parents. The interactions between these two possibly competing reasons for moving to any particular place can lead to spurious inferences about the impacts of school characteristics or parent's working on their children's development. By modeling the parents' choices of places of residence and their labor market behaviors (over 15,000 school districts), Dr. Mroz hopes to mitigate many of the potential biases that could have affected earlier studies.

The new Palmetto cluster enables Dr. Mroz’s research team to investigate how changes in school characteristics, arising from families moving across the more than 15,000 different school districts in the U.S., and changes in parents work behaviors affect the developmental outcomes of young children over time. In earlier work on a smaller scale computer cluster, the Mroz team was only able to examine a subset of these issues for much less informative sets of school characteristics and labor market opportunities. The detailed forms of dynamic economic models that they are now able to analyze will help to define the new state of the art in computational microeconomic models.

Protein Function and Evolution in Biological Systems

Brian Dominy, Assistant Professor of Chemistry at Clemson University, uses the Palmetto cluster to simulate the motion of various proteins and nucleic acid sequences in an effort to better understand how these molecules function and evolve in biological systems. First, by validating their dynamics calculations against experimental FRET analyses, descriptions of macromolecular dynamics can be derived that are simultaneously consistent with and more detailed than experimental measurements. Next, calculations of flexibility and changes in flexibility due to mutation can be correlated to experimentally determined changes in activity, giving a detailed atomic picture of the known relationship between dynamics and function. Further, molecular dynamics studies generate a statistical mechanical ensemble of conformational states from which thermodynamic properties (such as binding free energy) can be calculated and related to biomolecular function. In summary, conformational dynamic and thermodynamic properties of biomolecules are explored in order to better understand their function and to better understand how these molecules evolve.

Wireless Communications Research

Dan Noneaker, Associate Professor of Electrical and Computer Engineering at Clemson University, depends on the Palmetto cluster in his wireless communications research. Monte Carlo simulations based on high-fidelity models are used in research on cross-layer protocol design for mobile ad hoc packet radio networks. Novel computational-electromagnetic numerical-analysis algorithms are also used for E&M characterization in problems concerning antenna design, E&M penetration, and electromagnetic compatibility.

The capabilities of the new cluster allow consideration of wireless communication networks of a larger size using models of higher fidelity than was previously possible.

Manufacturing and Scheduling Optimization

Mary Beth Kurz, Assistant Professor of Industrial Engineering, has found the campus Condor computational pool invaluable for her research in manufacturing and scheduling optimization. Dr. Kurz employs genetic algorithms for large scale optimization in manufacturing and scheduling applications. By running hundreds of thousands of algorithmic processing jobs, Dr. Kurz helps manufacturers increase the throughput of their production processes.

Effect of Salt Concentration on the Binding Free Energy

Emil Alexov, Associate Professor of Physics and Astronomy at Clemson University, has performed a study to investigate the effect of salt concentration on the electrostatic component of the binding free energy based on a large set of 1482 protein-protein complexes, a task that had never been done before. Minimizing 2617 protein-protein complexes, some of which were larger than 50,000 atoms, wass quite a challenge from a computational point of view. Using Condor, the minimization of all structures was completed in less than two months.

Evolutionary Relationships of Genes

Bradley Rauh, Research Associate of Genetics and Biochemestry at Clemson University, studies the phylogenetic analysis of gene sequences describes the evolutionary relationships among those genes. These relationships are represented as a branching diagram, or tree. A phylogenetic tree together with other analyses, as gene expression studies, can provide information about gene function.

His group took advantage of the computational power of Palmetto Cluster to perform bootstrapping analysis of a maximum likelihood tree of 95 MIKC^C-type MADS-box genes from three different plant species (Arabidopsis thaliana, Populus trichocarpa and Prunus persica) using PAUP^* 4.0b10 software. These genes are generally associated with flower morphology and development in plants. The analysis exceeded the capacity of desktop computers due to the type of analysis performed (maximum likelihood) and the elevated number of genes included. However, the different replicates of the bootstrap analysis were run in parallel and finished in a short period using Palmetto Cluster.

Defect Detection in Carbon-Carbon Composites

James B. von Oehsen, Cyberinfrastructure Technology Integration Director of Computational Science at Clemson University, uses parallel finite element code to simulate the flow-microstructure relationship of mesophase pitch-based fluids that are used as precursors to produce carbon-carbon composites (such as those used in aircraft brakes and space-shuttle nose-cones). The simulation results obtained are for discotic liquid-crystalline materials using tensor-order constitutive equations. He is concerned with capturing the formation and evolution of defects that occur in these types of materials. The palmetto cluster has enabled him to run several simulations simultaneously to achieve solutions pertaining to different process conditions.