Utilizing a machine-learning algorithm, MIT researchers have identified a strong new antibiotic compound. In laboratory tests, the drug killed most of the world’s most problematic disease-inflicting bacteria, together with some strains that are immune to all known antibiotics. It also cleared diseases in two different mouse models.
The computer model, which might display over a hundred million chemical compounds in a matter of days, is meant to pick out potential antibiotics that kill bacteria utilizing different mechanisms than those of present drugs.
In their new study, the researchers recognized several different promising antibiotic candidates, which they plan to test further.
They believe the model may be used to create new medicines, based on what it has learned about chemical structures that allow drugs to kill microorganisms.
Barzilay and Collins, who are faculty co-heads for MIT’s Abdul Latif Jameel Clinic for Machine Learning in Health, are the lead authors of the paper, which comes today in Cell. The first author of the study is Jonathan Stokes, a postdoc at MIT and the Broad Institute of Harvard.
Over the past years, only a few new antibiotics have been created, and most of those newly cleared antibiotics are slightly different variants of existing medicines.
Current methods for testing new antibiotics are often prohibitively pricey, require a significant time funding, and are usually restricted to a narrow spectrum of chemical diversity.