Artificial Intelligence has found the most powerful antibiotic

Using a special machine learning algorithm, American scientists were able to identify a substance that can cope with many drug-resistant bacteria, including the causative agent of tuberculosis. It has already proven effective by curing laboratory drug-resistant mice. After 3 days of searching, the algorithm gave scientists a list of another 23 compounds, the verification of which has not yet been completed.

Antibiotics have given mankind the opportunity to treat a number of dangerous diseases. However, bacteria quickly develop resistance, so many of the existing drugs are already ineffective. Meanwhile, the ability to create new antibiotics is limited.

In search of new antibacterial drugs, researchers from the Massachusetts Institute of Technology turned to the help of artificial intelligence. The team, which Guardian talks about, fed the deep learning algorithm information on the atomic and molecular structure of 2500 natural and synthesized antibiotics, as well as their effectiveness against Escherichia coli Escherichia coli.

Researchers then asked the program to analyze a library of 6,000 compounds tested for the treatment of various diseases. The task of the algorithm was to select molecules that can effectively fight bacteria, but which are not similar in structure to existing antibiotics. This should have increased the chances of finding compounds that destroy microorganisms in fundamentally new ways.

In just a few hours, the algorithm identified several candidate substances. One of them, the failed medicine for diabetes (showed poor preclinical results, but the developers did not test it as an antibiotic), looked especially promising. Scientists named it chalitsin - in honor of the HAL artificial intelligence from the film "Space Odyssey of 2001". In structure, this compound was not similar to any known antibiotic.

Experiments have confirmed that chalicin can kill a number of drug-resistant microorganisms, including the tuberculosis pathogen Mycobacterium tuberculosis and the latest antibiotic-resistant strains of the bacteria from the Enterobacteriaceae family. In addition, the drug cured laboratory drug mice from Clostridium difficile and Acinetobacter baumannii with multidrug resistance.

“I think this is one of the most powerful antibiotics that have been discovered to date,” said James Collins, a MIT bioengineer. “It has a very high activity against a wide range of antibiotic-resistant pathogens.” Moreover, for human cells, it is non-toxic.

After the discovery of chalicin, the team loaded data on 107 million chemical compounds into the algorithm. Three days later, the program issued a list of 23 potential antibiotics, eight of which proved to be effective in experiments, and two more may be especially strong. The authors note that without the use of artificial intelligence, it would be impossible to test such a quantity of substances.

Now researchers intend to use the algorithm to detect molecules that are effective against specific bacteria. When taking such antibiotics, it will be possible to avoid the death of beneficial symbiotic microorganisms. Another task that researchers also intend to try to solve using this algorithm is the development of new antibiotics from scratch.

Creating new antibiotics is not the only approach to fighting bacteria. For example, Australian scientists have found that a combination of approved drugs can kill bacteria inside biofilms. Conventional antibiotics cannot do this.

Post a Comment