Embedding chemical space for generative discovery of polypharmacology drugs. Credit: Brenton P. Munson et al.
UC San Diego scientists have developed a novel machine-learning algorithm to simulate the chemistry involved in early-phase drug discovery. The algorithm could aid future drug discovery research, providing a more streamlined approach to discovering groundbreaking treatments.
Typical drug discovery research involves thousands of experiments to select and optimize a candidate treatment. With the new AI platform developed by the researchers, the same results could be achieved in a fraction of the time. Published in Nature Communications, the tool was successfully used to synthesize 32 new cancer drug candidates.
"A few years ago, AI was a dirty word in the pharmaceutical industry, but now the trend is definitely the opposite, with biotech startups finding it difficult to raise funds without addressing AI in their business plan," said Trey Ideker, professor in the Department of Medicine at UC San Diego School of Medicine and adjunct professor of bioengineering and computer science at the UC San Diego Jacobs School of Engineering. "AI-guided drug discovery has become a very active area in industry, but unlike the methods being developed in companies, we're making our technology open source and accessible to anybody who wants to use it."
The platform, called POLYGON was trained on a database consisting of more than 1 million known bioactive molecules including information about their chemical properties and protein target interactions. By learning patterns from this database, POLYGON can generate chemical formulas for new candidate drugs that lack certain undesirable properties such as protein inhibition. Additionally, POLYGON is capable of identifying molecules with multiple targets.
"Just like AI is now very good at generating original drawings and pictures, such as creating pictures of human faces based off desired properties like age or sex, POLYGON is able to generate original molecular compounds based off of desired chemical properties," said Ideker. "In this case, instead of telling the AI how old we want our face to look, we're telling it how we want our future drug to interact with disease proteins."
To test POLYGON, the team used it to generate hundreds of potential drugs to target various cancer-related proteins. Of those generated, the team synthesized 32 of them that were the most likely to interact with MEK1 and mTOR proteins, a promising target for future combination therapy. The drugs synthesized showed significant activity against MEK1 and mTOR, but a few had off-target reactions with other proteins.
The team is optimistic about the possibilities presented by AI in the field of drug discovery. While AI is unlikely to completely replace humans and the need for hands-on chemistry, it could streamline the discovery process leading to novel treatments in the future.
"Once you have the candidate drugs, you still need to do all the other chemistry it takes to refine those options into a single, effective treatment," said Ideker. "We can't and shouldn't try to eliminate human expertise from the drug discovery pipeline, but what we can do is shorten a few steps of the process."