Dark Mode Light Mode

AI is a new antibiotic target for intestinal bacteria | MIT news

Spread the love


In patients with inflammatory bowel disease, antibiotics can be double -edged knives. A wide spectrum drug, which is often prescribed for intestinal flare up, can kill harmful microorganisms and useful microorganisms, and sometimes symptoms may worsen over time. When fighting intestinal inflammation, you always don’t want to bring sleds to the sword fight.

MIT’s computer science and artificial intelligence research institute (CSAIL) researchers and MCMaster University Identified new compounds You need a more target approach. The molecule, called Eternololin, has left the rest of the microorganisms as it suppresses bacteria groups related to Crohn’s bottle flare up. The team used the creation AI model to map the compounds of the compound. In general, it takes a few years, but there is a process accelerated for several months.

“This discovery is related to the central challenge of antibiotic development. A new thesis on the workMcMaster’s Biochemistry and Biological Science Professor and MIT’s ABDUL LATIF JAMEEL Clinic’s Health Research affiliates. “The problem is not to find a molecule that kills bacteria on a plate. We have been able to do so for a long time. A big obstacle is to find out what he acts inside the bacteria. Without a detailed understanding, these early stages of antibiotics cannot be developed to patients safe and effective.”

Enpollo Rollin is a progress for precision antibiotics. Treatment designed to exclude bacteria only causes problems. In mouse models of inflammation such as Kron, the drug They appeared coldIntelligent bacteria, which can worsen flare, and most other microorganisms do not reach. The minister’s mouse recovered faster and more healthy microorganisms than the general antibiotic, Vancomycin.

When the mechanism of action of the drug is fixed, molecular targets that combine inside bacterial cells usually require many years of difficult experiments. Stokes’ laboratory found Enrolo Lin using a high -through -handed screening approach, but it would have been a bottleneck to determine the goal. Here the team headed to Difdock, a generated AI model developed by MIT PHD student Gabriele Corso and MIT professor Regina Barzilay.

Diffdock is designed to predict how small molecules are suitable for the combination pocket of protein, and are notorious for rescue biology. Existing docking algorithms search in the direction as possible using score rules, often of noisy results. DIFFDOCK instead of docked by probability loan reasoning, framework docks: diffusion models repeatedly improve speculation until convergent in the most likely binding mode.

Barzilay, who co -reads Jameel Clinic, said, “In just a few minutes, the anti -polo rollin binds to a protein complex called LoLCDE.” This was a very specific lead. Rather than replacing the experiment, it was a lead to guide the experiment. “

Then the group of Stokes put the prediction in the test. The Difdock prediction was used as an experimental GP, and they first evolved Enterololin-resistant mutation. E. coli In the laboratory, the change in the DNA of the mutant body was found to be mapped to LoLCDE. In addition, when exposed to the drug, RNA sequencing was performed to check the bacterial gene that was turned on or off by the bacterial gene, and CRISPR was used to selectively knock on the expression of the expected target. All of these laboratory experiments were destroyed in the path associated with lipoprotein transportation.

Stokes says, “It’s time to believe that you have found something when you see the calculation model and wet rap data pointing to the same mechanism.”

For Barzilay, this project emphasizes changes in the way AI is used in life science. “The use of many AIs in drug discovery is to identify new molecules that can be activated by searching for chemical spaces,” she said. “What we see here is that AI can provide an important mechanical explanation for moving molecules through the development pipeline.”

These differences are important because the mechanism of action mechanism is often the main speed limit stage of drug development. Traditional approaches can take more than 18 months to two years and can take millions of dollars. In this case, the MIT -MCMARTE team cut the timeline to about 6 months and cut it as part of the cost.

Enterololin is still in the early stages of development, but the translation is already in progress. Stoked BIO, a spin -out company in Stokes, optimizes the characteristics for licensing and potential human use of this compound. Initial research also explores the derivatives of the molecules for other resistant pathogens. Klebsiela pneumoniae. If everything goes well, clinical trials can begin in the next few years.

Researchers also have a wider impact. Narrow spectrum antibiotics have been visited for a long time as a way of treating infections without damage to microorganisms, but it was difficult to find and verify. AI tools, such as DiffDock, can create that process more practically, allowing new generation of target antibacterial agents quickly.

For patients with kroron and other inflammatory bowls, the prospects of drugs that reduce symptoms without unstable microbial cluster can mean improvement in which the quality of life is meaningful. And in a larger figure, precision antibiotics can help to solve an increasing threat of antibacterial resistance.

“I can think about the mechanism of behavioral explanation as I can do it faster with this compound as well as this compound, as well as this compound, as well as AI, human intuition and laboratory experiments.” This is likely to change how to approach the discovery of drugs for many diseases as well as crons. “

“One of the biggest challenges for our health is the antibacterial resistant bacteria that evacuate our best antibiotics,” “AI is an important tool for this fight with bacteria.

CORSO, BARZILAY and STOKES are Denise B. Catacutan, Vian Tran, Jeremie Alexander, YEGANEH YOUSEFI, Megan Tu, Stewart MCLELLAN and Dominique I wrote a paper with Jakob Magolan, Michael Surette, Eric Brown and Brian Comebes. Their research was partially supported by the Weston Family Foundation. David Braley Center for discovery of antibiotics; Canadian Health Research Institute; Canadian Natural Science and Engineering Research Committee; M. and M. Heersink; Canadian research institute for health research; Ontario Graduate School of Scholarship Award; Jameel clinic; And US defense threats reduced agencies discovers medical response to new and emerging threats.

The researchers posted sequencing data in the open repository and publicly presented the Difdock-L code in GitHub.



Source link

Keep Up to Date with the Most Important News

By pressing the Subscribe button, you confirm that you have read and are agreeing to our Privacy Policy and Terms of Use
Add a comment Add a comment

Leave a Reply

Previous Post

Aida Ho is 48th in the new mental health status report

Next Post

A case of an Indian woman in a lawsuit against Trump for H-1B fees: 'You will have to leave her… '