MIT Researchers Discover Miracle Antibiotic Using Artificial Intelligence

Intan Rakhmayanti Dewi, CNBC Indonesia

Tech

Monday, 01/15/2024 19:10 IWST

Photo: REUTERS/RULA ROUHANA

Jakarta, CNBC Indonesia – Researchers discovered a group of compounds that can kill drug-resistant bacteria, which cause more than 10,000 deaths in the United States each year.

The miracle drug was discovered using a type of artificial intelligence (AI) with deep learning by MIT researchers.

In a study published in Nature, researchers showed this compound could kill methicillin-resistant Staphylococcus aureus (MRSA) growing in laboratory dishes and in two mouse models of MRSA infection. The compound also shows very low toxicity to human cells, making it a good drug candidate.

The main innovation of the new study is that the researchers were also able to find out what kind of information the deep learning model used to make predictions about antibiotic potency.

This knowledge could help researchers design additional drugs that might work better than the drugs identified in the model.

“The insight here is that we can look at what the model learns to make predictions that a particular molecule will make a good antibiotic,” said James Collins professor in MIT’s Institute of Medical Engineering and Sciences (IMES) and Department of Biological Engineering.

“Our work provides a time-saving, resource-saving and mechanistically insightful framework, from a chemical structure standpoint, in a way that we have not done to date,” he added.

Felix Wong, a postdoc at IMES and the Broad Institute of MIT and Harvard, and Erica Zheng, a former Harvard Medical School graduate student are the lead authors of the study, which is part of the Antibiotic-AI Project at MIT.

The mission of the project led by Collins is to discover new classes of antibiotics against seven types of deadly bacteria over seven years.

The researchers trained a deep learning model using a substantially expanded dataset. They generated this training data by testing about 39,000 compounds for antibiotic activity against MRSA. Then they input this data, plus information about the chemical structure of the compound into the model.

“You can basically represent any molecule as a chemical structure, and you can also tell the model whether that chemical structure is antibacterial or not,” Wong said.

“The model is trained with many of the same examples. If you then feed it new molecules, new atomic arrangements and bonds, it can tell you the probability that the compound is predicted to be antibacterial.”

To find out how the model makes its predictions, the researchers adapted an algorithm known as Monte Carlo tree search, which has been used to help build other deep learning models, such as AlphaGo.

This search algorithm allows the model to generate not only estimates of the antimicrobial activity of each molecule, but also predictions of which molecular substructures are most likely responsible for that activity.

Watch the video below: