Tue, Nov 18, 2003: RAPTOR Ranks Top in International Automated Protein Structure Prediction Competition
BSI is delighted to announce that a new protein structure prediction program RAPTOR (Rapid Protein Threading predictOR), jointly developed by University of Waterloo computer science Ph.D. student Jinbo Xu, and UW professor and CEO of BSI, Ming Li, in collaboration with Dr. Ying Xu's group at Oak Ridge National Lab, ranked top in the renowned Critical Assessment of Fully Automated Structure Prediction (CAFASP3/CASP5) competition in December 2002.
The competitions were conducted entirely via the internet. The goal of CAFASP/CASP is to evaluate the performance of fully automatic structure prediction servers available to the community. the RAPTOR program developed by Xu and Li was ranked No. 1 among automatic non-meta prediction programs for 3D (three-dimensional) protein structure prediction in the Fold Recognition category. Moreover, RAPTOR was voted by peers to be the most interesting new method at CASP5 (voting results at http://forcasp.org).
"RAPTOR's No. 1 ranking is a very high achievement since protein structure prediction is a difficult and very important field, and many famous people have been working on this for all their lives," said Li. Over the years, more than 100 teams have participated in the CASP/CAFASP competitions. "No Canadian team has ever achieved anything within the top 10," Li said. He added that while RAPTOR supplants all existing automatic protein predictions, it is still not solving the problem fully. "This is important work because protein structure prediction is an extremely hot area, and key to the success of worldwide proteomics efforts."
Li, who also holds UW's Canada Research Chair in Bioinformatics, explained that the proteomics projects in the so-called post-genome era focus on understanding of proteins. "Determining the three dimensional structures of proteins is a key step from genes to drugs."
Li said that some experts go as far as saying "any breakthrough in the high-throughput protein folding would cause revolutionary change" in ways people do research and in pharmaceutical industry. "Jinbo's work is certainly one step towards this direction." At present, the low-throughput method used in the wet labs could take one scientist as long as six months to determine the structure of one protein, deploying X-ray crystallography or NMR (nuclear magnetic resonance) technologies.
"Therefore, automated computer prediction is urgently needed but progress has been very slow," Li said, noting that hundreds of top research groups in the world have been working on the challenge for over a decade.
BSI's PROSPECT Pro was the top-ranking threading-based program in CASP 4. While the algorithmic approaches each software takes differ, Li believed that together, the two software can complement each other and provide even more accurate and reliable consensus results. BSI is the exclusive commerical distributor of RAPTOR.
