Duke software dramatically speeds enz...( DURHAM N.C. -- A Duke University-led...)

DURHAM, N.C. -- A Duke University-led team has brought powerful software to the never-ending arms race between antibiotics and germs. Working together, computer scientists and biochemists have developed and laboratory-tested a computer program that can show experimentalists how to change the machinery that bacteria use to make natural antibiotics.

The program -- a set of computer rules known as the algorithm "K*" (pronounced K Star)-- is able to sort through all the possible shapes and changes of a key enzyme that produces a natural antibiotic called gramicidin S, said Bruce Donald, Duke's William and Sue Gross Professor of Computer Science and Biochemistry. The new technique might pave the way toward more automated redesign of old drugs to foil drug-resistance in germs.

"It really excites us that we can redesign enzymes on a computer, make them in the laboratory and have them work as planned," said Donald, who leads the extended research effort and is corresponding author of a new report to be published online the week of Feb 16 in the research journal Proceedings of the National Academy of Sciences (PNAS).

The work was funded by a grant from the National Institutes of Health.

The search for new antibiotics usually begins by directly modifying existing compounds. But his group instead predicted mutations to enzymes from an antibiotic-making microbe, using K* to search faster and cheaper for the best designs.

"It is essentially a new pathway to make novel antibiotics," Donald said. "There are many possible changes you can make to a protein, but the algorithm can test out orders of magnitude more variations than laboratory experiments alone."

Other protein design algorithms have been proposed, and some of those even attempt to account for the way key parts of real proteins move around in three dimensions. But Donald said the latest version of K* lets protein backbones and side chains wiggle
'/>"/>