A Web-Accessible Protein Structure Prediction Pipeline

Abstract

Proteins are the molecular basis of nearly all structural, catalytic, sensory, and regulatory functions in living organisms. The biological function of a protein is inextricably linked to its three-dimensional (3D) atomic structure. Traditional structure determination methods, such as X-ray and nuclear magnetic resonance techniques, are time-consuming, expensive, and infeasible for the millions of proteins that have been sequenced so far from various organisms. Alternatively, computational structure prediction methods provide a faster and more cost-effective, albeit approximate, alternative to experimental structure determination. We present a high-throughput protein structure prediction pipeline (dubbed “PSPP”), which given input protein sequences infers their 3D atomic structures. The pipeline was designed to be used with high performance computing clusters and to scale with the number of processors. The pipeline encompasses a core Perl module, a parallel job manager, and a Web browser graphical user interface accessible at our Website (www.bhsai.org). The software is currently installed at the Department of Defense (DoD) Maui High Performance Computing Center, and it is available for download along with its associated databases from our site. Currently, DoD scientists are using the pipeline in basic science and drug and vaccine development projects.