Professor David Baker and his colleagues came in first this year, unimaginably far ahead of a distinguished pack of more than 100 research groups, in the fourth annual international Computer Assisted Structure Prediction Competition ("CASP4"). This event is a veritable Molecular Olympics, with more than 100 research groups from around the world attempting to predict three-dimensional protein structures from one-dimensional amino acid sequence information. The crystal structures of the proteins are known, but not disclosed until the structural predictions have been submitted to the competition office.

Successful structure prediction has been the proverbial Holy Grail of structural genomics. The computational technique developed by the Baker group, and known as ROSETTA, represents a giant step toward interpreting the nearly complete human genome sequence, as well as applying this new knowledge to drug design, genetics, and health care.

The human genome sequence would be nearly useless if we could not understand the structures of the proteins it encodes. Although X-ray crystallography and nuclear magnetic resonance (NMR) are the only methods that can determine protein structures precisely, approximate methods are invaluable because there is neither time, nor funding, nor personnel, nor luck enough to express, purify, and determine the structure of every protein experimentally. Thus computational techniques like ROSETTA, or variations on it, will remain one of the major routes from genome sequence to protein structure for many years to come.

For additional information, see http://www.hhmi.org/news/baker.html and http://predictioncenter.llnl.gov/casp4.