Scientists at UC San Diego have developed a machine learning algorithm to mimic the time-consuming chemical research experiments that occur at the start of drug discovery, dramatically simplifying the process and opening the door to unprecedented cancer treatments.
The study was conducted by researchers at the University of California, San Diego, USA and was published on May 6, 2024. Journal of Nature Communication(Nature Communications) and wrote about it EurekAlert website.
Thousands of trials are being conducted to develop candidate drugs as treatments, and a new artificial intelligence platform can deliver the same results in a shorter period of time. The researchers used the new tool to design 32 new drug candidates for cancer treatment.
This technology is part of a new but growing trend in pharmaceutical science to use artificial intelligence to improve the drug discovery and development process.
“AI-guided drug discovery has become a very active field in the industry,” said lead author Prof. Trey Edeker, professor of internal medicine at the UC San Diego School of Medicine and assistant professor of bioengineering and computer science at the UC San Diego Jacobs School of Engineering. In contrast, we make our technology open source and available to anyone who wants to use it.”
Polygon
The new platform, called POLYGON, is unique among AI tools designed for drug discovery in that it can identify molecules that target multiple proteins, whereas current drug discovery protocols currently prioritize treatments that work on a single target. Multi-targeted drugs are of great interest to clinicians and scientists because they offer the same benefits as combination therapy with more than one drug.
Dr. Edeker pointed out that it takes years and millions of dollars to discover and develop a new drug, especially if we are talking about a drug that has multiple targets. He explained that serendipity is behind the discovery of some of the most versatile drugs we have, but this new technology will remove serendipity from the equation and help usher in a new generation of precision medicine.
The researchers trained Polygon on a database that contains more than a million biologically active molecules and contains detailed information about their chemical properties and interactions.
By learning from patterns in the database, Polygon can generate original chemical formulas for new drug candidates that may have certain properties, such as the ability to inhibit specific proteins.
Explaining the programming work, Edeker said: “Artificial intelligence is now very good at creating original drawings and images, such as creating images of human faces based on characteristics you ask for, such as age or gender, which Polygon can do. To create original molecular compounds based on desired chemical properties. In our case, he points out, instead of telling the AI how old our face should be, we’re telling our future drug how to interact with disease proteins.
To test Polygon, researchers used it to generate hundreds of drug candidates that target different pairs of cancer-related proteins. Among these molecules, the researchers combined 32 molecules with the MEK1 and MTOR proteins, a pair of cellular signaling proteins that are a promising target for cancer therapy, and inhibiting one alone is not enough.
Medicine for cancer
The researchers found that their drugs had significant activity against MEK1 and MTOR, but some off-target interactions with other proteins. This suggests that one or more of the drugs identified by Polygon could target both proteins as cancer therapeutics, giving human chemists a menu of options that can be fine-tuned.
“Once you have drug candidates, you still have to do all the other chemical processes necessary to develop these options into an effective treatment,” Edeker said. “We cannot and should not try to exclude human expertise from the drug discovery pathway, but what we can do is cut some steps out of the process,” he warned.
Edgar expressed his hope; Emphasizing how this concept will evolve over the next decade, whether in academic circles or in the private sector, is very exciting, noting that the possibilities are almost endless.
