Rosetta ab initio protein folding
EMSL Project ID
5093
Abstract
We propose developing tools and scheduling techniques to make use of the unutilized compute cycles on the HP supercomputer in the Molecular Science Computing Facility to predict the structure of several thousand proteins of unknown function from several genomes of interest staff at the Institute for Systems Biology and the Department of Energy. The method we propose to predict protein structure in the absence of sequence-homology to proteins of known structure is Rosetta, an ab initio protein structure prediction program developed at the University of Washington. Predicted protein structures will then be used in combination with other bioinformatics tools, and global experimental observations, to uncover the functional roles of these proteins of unknown function. The Rosetta structure prediction procedure is a computationally intensive Monte Carlo procedure for finding low-energy conformations for a given protein sequence. The annotations resulting from these structure predictions will be made publicly available/accessible upon publication. We have only recently shown that ab initio structure predictions can be useful in uncovering general function for proteins of unknown function. This development is well timed with respect to other novel developments in bioinformatics and systemsbiology (prediction of protein interaction networks and methods for collecting genome-wide observations), and thus the results from this work will be central to a larger data-integration, visualization and annotation effort currently underway at the Institute for Systems Biology. By making minor modifications to the scheduling system on the HP supercomputer we believe it will be possible to run these calculations in the unutilized cycles, thereby increasing the utilization rate of the system. One tangible short term scientific goal for the project is to use the system to perform Rosetta predictions for all unknown function genes in the newly sequenced genome Haloarcula marismortui, which will be the first application of ab initio structure prediction to the initial annotation of a complete genome.
Project Details
Project type
Exploratory Research
Start Date
2003-10-16
End Date
2004-10-21
Status
Closed
Released Data Link
Team
Principal Investigator