William B. Gleason, Fellow

Applications of Advanced Computation and Digital Simulation to Problems of Biomedical Relevance

The non-redundant PIR file (a protein sequence file downloaded March 2, 2001) is used as a test database to search for five peptide sequences derived from MS. “Wall-clock” time drops from approximately 120 seconds on a single processor to approximately 20 seconds using 8 Pentium processors operating at 700 MHz.

The Gleason group is continuing to explore parallel processing strategies of potential widespread utility in mass spectrometry. These include rapid “on-the-fly” analyses that allow for immediate identification of peptide analytes or, in the case of a failure to identify, immediate interactive modification of instrumental parameters for further experimentation. Such an application is particularly appropriate for testing on the Supercomputing Institute’s Netfinity LINUX cluster, where the group can test their ideas and determine the appropriate number of processors to use for a given application. An example of the use of a “naïve” parallelization strategy is shown in the figure (next page). This indicates that a small laboratory LINUX cluster of perhaps eight processors (running at about 2 GHz) would be suitable for an “on the fly” application. Off-line analyses still remain of great importance, particularly those that involve large databases such as the ever-expanding genomic databases for humans as well as other species. Here, parallel processing using the Supercomputing Institute’s high-performance computers with large memory and disk capabilities is attractive.

Of continuing interest to the group is the fibroblast growth factor (FGF) family of proteins, now numbering more than twenty distinct proteins, not counting the observed phenomenon of alternative splicing, which produces additional variants of each sequence. FGF family members have diverse biological roles ranging from control of neurite outgrowth in normal development to involvement in the wound-healing process. FGF members share a common property, i.e. they bind heparin, a sulfated polysaccharide best known for its anticoagulation properties. The group is interested in studying the interaction of proteins with sulfated carbohydrate polymers related to heparin. Although there are crystal structures available for some family members many important FGFs have available only sequence information. Thus these researchers are very interested in the use of homology modeling to construct three-dimensional models for FGFs using known crystal (or nuclear magnetic resonance) structures as templates. To that end they have constructed models for FGF-7 (whose three-dimensional structure is known) as well as FGF-13 for which no crystal structure is currently available.

The Gleason group continues to take a “tools approach” to their work in structural bioinformatics, dating back to the Protein Analysis Package (PAP) developed at the Supercomputing Institute in the early 90s. Thus Eric Johnson and Bill Gleason were invited to present a tutorial/workshop on the subject at the O’Reilly Bioinformatics meeting in early 2003. The highlight of the presentation was a series of movies made using the AMIRA package at the Supercomputing Institute that depicted a molecular dynamics simulation of heparin, an FGF family member, and an extensive illustration of the structural features of heparin important for its biological function.