Researchers at the Broad Institute of MIT and Harvard, Integrated Biosciences, Wyss Institute for Biologically Inspired Engineering, and Leibniz Institute of Polymer Research have discovered a class of structurally novel antibiotics.This is the first time a new class of antibiotics has been discovered in the past 60 years and the first using an artificial intelligence-driven platform built around explainable deep learning.
This peer-reviewed paper was published in the journal Nature today (December 20), titled "Discovery of a structural class of antibiotics with explainable deep learning" (Discovery of structural class of antibiotics with explainable deep learning). The paper was published by a Co-authored by a team of 21 researchers, led by Dr. Felix Wong, co-founder of Integrated Biosciences, and Dr. James J. Collins, Termeer Professor of Medical Engineering and Science at MIT and founding chairman of the Integrated Biosciences Scientific Advisory Board.
Other collaborators include researchers at MIT, the Broad Institute of MIT and Harvard, the Wyss Institute for Biologically Inspired Engineering, and the Leibniz Institute of Polymer Research in Dresden, Germany.
In the study, the researchers virtually screened more than 12 million candidate compounds to identify this new class of antibiotics that showed potential to address the problem of antibiotic resistance.
In this groundbreaking approach, the research team trained a deep learning model on experimentally generated data to predict the antibiotic activity and toxicity of any compound. Drawing inspiration from artificial intelligence used in other fields, such as DeepMind's AlphaGo game technology, the authors designed new models to explain which parts of the molecule are important for antibiotic activity.
The results identified a new class of antibiotics with potent activity against multidrug-resistant pathogens. In a series of experiments, researchers tested a candidate antibiotic in a mouse model of MRSA infection and found that it was effective both locally and systemically, suggesting the compound is suitable for further development as a treatment for severe and sepsis-related bacterial infections.
"The discovery of this new class of antibiotics is a breakthrough achievement that demonstrates the unique ability of artificial intelligence and explainable deep learning to catalyze drug discovery," said Dr. Wong. "Our work exposes several high-energy models that can accurately predict antibiotic activity and toxicity. Importantly, this is one of the first demonstrations that a deep learning model can explain its predictions, with direct and profound implications for how drug discovery is conducted and how artificial intelligence can be used to efficiently discover new drugs."
"This is an important validation of how important the integration of artificial intelligence with interpretable deep learning is to overcome some of the toughest challenges in medicine, in this case antibiotic resistance," said Dr. Collins. "Building on these confirmatory studies and similar approaches, the IntegratedBiosciences team is poised to further accelerate the integration of synthetic biology with a deep understanding of cellular stress responses to address the significant unmet need for new treatments for age-related diseases."
"An important implication of this study is that deep learning models in drug discovery can, and in many cases should be solvable," said Dr. Satotaka Omori, founding member and head of aging biology at Integrated Biosciences and a contributing author on the paper. While AI continues to have an impact, it is also limited by the many black-box models that are commonly used to obfuscate the underlying decision-making processes. By opening these black boxes, we aim to create more generalizable insights that may be more useful in accelerating the use and development of next-generation drug discovery methods."
Alicia Li, a research associate at Integrated Biosciences and a contributing author on the publication, added: "It's really exciting to see that we can demonstrate a new way to predict how well a compound will work as an antibiotic, the likelihood of the compound progressing in phase 1 trials, and whether the compound is one of potentially many other members of a new class of drugs."
Integrated Biosciences has established a research system that, in addition to this newly published "Nature" paper, also includes a "Nature-Aging" paper published in May, which shows how to use artificial intelligence to discover new senolytics, that is, anti-aging compounds that selectively eliminate aging "zombie" cells. These compounds hold promise for treating diseases associated with aging, such as fibrosis, inflammation, and cancer.
A Cell Systems paper published in July demonstrated a synthetic biology-based platform that allows humans to control aging-related stress responses, thereby accelerating drug screening for aging.
Reference: "Discovering a class of antibiotic structures using explainable deep learning", December 20, 2023, "Nature".
DOI:10.1038/s41586-023-06887-8
Compiled source: ScitechDaily