Biomedical Engineering Reference
In-Depth Information
Endnote
For researchers who focus on modeling the 3D structure of proteins, the biannual meeting of the
Critical Assessment of Techniques for Protein Structure Prediction (CASP) is a moment of truth.
Participants, whose future funding often rests on the results of their efforts, are challenged with the
task of using their computer algorithms to predict the precise 3D structure of an array of relatively
small proteins. Results are judged by comparing predicted structures with those determined
experimentally through NMR or X-ray crystallography methods. Before the competition was limited to
fully automated means of structure prediction, one of the top performers at the conference wasn't an
algorithm running on a supercomputer or Linux cluster, but a mere human—Alexey Murzin of the
Medical Research Council's Laboratory of Molecular Biology in Cambridge, England.
Murzin's predictions of protein structure, which are based on biological heuristics, cast such a doubt
over the need for computational methods that there was talk of cutting government funding to the
computer laboratories involved in structure prediction. Even though Murzin is now barred from
officially competing in the current CASP meetings, his heuristic methods consistently compare
favorably with the best computational solutions.
The human-machine competition in bioinformatics, like the competition between Garry Kasparov and
a chess-playing computer in 1997, is likely to eventually be won by the machine, if for no other
reason than the lack of human experts to predict the structure of the thousands of proteins
discovered every year. However, even as the price of computing power drops, Murzin's performance
suggests that simply adding faster processors isn't the best solution, and that a combination of ab
initio and heuristic methods will eventually provide the most accurate, consistent, and affordable
predictions of protein structure.
