A computer program developed at Duke has become the latest weapon in the war on drug-resistant bacteria.

Researchers at Duke and the University of Connecticut have published findings that open source software can predict how bacteria will become resistant to new drugs. The software analyzes the structure of the bacteria and determines how bacteria will change and adapt to new drugs. In lab tests, the software was able to predict how Methicillin-resistant Staphylococcus aureus bacteria would develop resistance to new drugs developed at the University of Connecticut. The software may now be used to improve the drug resistance of antibiotics before they undergo clinical trials.

“It gives us a little window onto evolution,” said Bruce Donald, James B. Duke professor of computer science and chemistry and co-author of the paper. “We can see what moves the bacterium could make—or will make—against new drugs that are in the pipeline."

Donald’s open source software—OSPREY—has been used previously to determine the structure of proteins and design new proteins. For this experiment, the software was used to determine how bacteria could modify their proteins in ways that would make them resistant to new drugs.

“The hypothesis is that in order to select resistance mutations, essentially the pathogen has to, on its own, solve a rather tricky protein design problem,” Donald said

Once the software predicted the possible ways in which bacteria might change to fight new drugs, those predictions were tested at the University of Connecticut by Amy Anderson, professor of medicinal chemistry. She is currently working on developing new antibiotics to fight drug-resistant MRSA bacteria. Anderson exposed bacteria to her new drugs and saw how the bacteria changed to develop resistance.

“We grew resistant bacteria and we sequenced the gene to find out which mutations it had. We found some of the same mutations which [the software] had predicted,” Anderson said. “The prediction was actually validated in the bacteria.”

Once researchers know how bacteria will adapt to a new drug, they can modify the drug to fight those adaptations. This saves a large amount of time, effort and money in the drug development process, since researchers don’t have to run costly experiments or even clinical trials to determine how bacteria will develop resistance, Anderson said.

She added that the techniques developed in this study could be introduced very soon in to the development process for new drugs.

“There’s no reason it couldn’t be used now for new antibiotics.” Anderson said.

In the future, both Anderson and Donald hope to expand their research in order to aid the development of drugs to fight other types of diseases.

Pablo Gainza-Cirauqui, a graduate student in Donald’s Lab, pointed out that the new techniques could be used against diseases which are harder to examine in the lab.

“There are other organisms that are very hard to grow in colonies, for example some viruses and some cancer cell lines,” he said. “We think it would very interesting if we were to apply this technology to those pathogens and diseases that develop resistance but which we cannot actually grow and test.”

Donald also noted that he hoped using the software to combat drug resistance would help to preserve the value of new drugs targeting widespread diseases such as AIDS. Currently, many newly-developed drugs become less effective soon after they are introduced, if measures are not taken to mitigate drug resistance.

"If while the clinical trials [for HIV drugs] get under way, [the software] might be able to predict ahead of time what kind of mutations the antigenic targets might evolve or select in order to evade our designed antibodies, that would be very exciting," Donald said.