Apr 6, 2020
For the past decade, the Collins Lab at MIT has been focused on using bioengineering principles to better understand antibiotics with the primary goal of discovering novel molecules that work effectively against bacterial pathogens.
On this episode, you’ll learn the following:
- What four primary mechanisms of antibiotic resistance are used by pathogens
- How AI can be used to identify certain features of molecules out of massive numbers of molecules and amounts of data
- Where Collins hopes to see his research and applications applied in the coming years
James J. Collins, Ph.D., is a professor of medical engineering and science at the Broad Institute of MIT and Harvard, and head of the Collins Lab at MIT. About one year ago, he teamed up with colleague Regina Barzilay, one of the world’s leading experts on applying artificial intelligence (AI) to healthcare.
The goal was to determine whether the power of AI could be used to address the challenge of antibiotic resistance and bacterial pathogens through the discovery of new antibiotics.
They began by putting together a training collection of over 2,500 molecules, including 1,700 FDA-approved drugs. This library was tested against E. coli in the lab to see which molecules might lead to inhibitory activity against the bug. Next, a deep neural network was trained using the data gathered and information about the structure of each molecule in the library.
The trained deep neural network was then applied to a drug repurposing library containing several thousand molecules that have already been developed or are in the process of being developed as drugs. The neural network was challenged to identify molecules that are predicted to be antibiotics but don’t look like any existing antibiotics: one molecule fit the criteria, and was named halocin. Halocin proved itself to be a potent novel antibiotic that worked against 35 out of 36 samples of multidrug-resistant, extensively drug resistant and pandrug-resistant pathogens from the CDC.
In addition to the details of this exciting discovery that could change health and medicine for the better, Collins discusses the most common mechanisms of bacterial resistance to antibiotics, why gram negative bacteria poses an extra challenge to the search for effective antibiotics, how AI could be used to identify features of molecules that make them amenable to gram negative bacterial uptake, the most useful strengths at the core of the AI technology being used in these capacities, the soon-to-be-launched Antibiotics AI Project, and so much more.
Tune in for the full conversation and learn more at collinslab.mit.edu.