The goal of a new drug screening method based on artificial intelligence and developed by researchers at the University of Central Florida is to speed up the expensive and time-consuming process of making life-saving medicines. The researchers identified promising drug candidates with up to 97% accuracy using a technique that models drug and target protein interactions using natural language processing techniques. Recent issues of the journal Briefings in Bioinformatics featured the results.
In the method, each protein binding site is represented by a word. Then, deep learning is used to find the features that control how the two parts interact.
According to study co-author Ozlem Garibay, an assistant professor in the UCF Department of Industrial Engineering and Management Systems, “With AI becoming more accessible, this has become something that AI can tackle.” You can experiment with a huge variety of protein and drug interactions to determine which ones are most likely to bind or not.
Their model, called AttentionSiteDTI, is the first that can be understood in terms of protein binding sites.
The work is important because it will help find important protein binding sites and figure out how they work, which is important for figuring out if a drug will work or not.
Researchers made a self-attention mechanism that makes the model learn which parts of the protein interact with the drug compounds. This way, they were able to make predictions that were on the cutting edge.
The mechanism’s ability to pay attention to itself works because it only pays attention to the most important parts of the protein.
The researchers used in-lab experiments to measure the binding interactions between various compounds and proteins, and then they compared the outcomes to those that their computational model had predicted. The experiments also included testing and validating drug compounds that would bind to a spike protein of the SARS-CoV2 virus because drugs to treat COVID are still of interest.
According to Garibay, the strong agreement between experimental findings and computational forecasts demonstrates AttentionSiteDTI’s potential to pre-screen potentially useful drug compounds and hastened the development of both new treatments and their repurposing.
Sudipta Seal, study co-author and chair of UCF’s Department of Materials Science and Engineering, states that “this high-impact research was only possible due to interdisciplinary collaboration between materials engineering and AI/ML and computer scientists to address COVID-related discovery.”
The lead author of the study, Mehdi Yazdani-Jahromi, is a doctoral student in UCF’s College of Engineering and Computer Science. He says the work is a first step toward a new way of pre-screening drugs.
According to Yazdani-Jahromi, this enables researchers to use AI to identify drugs more precisely and respond quickly to new diseases. The best binding site of a virus’s protein to concentrate on when developing a drug can also be found by the researchers using this method.
The next phase of his research will involve creating new medications using artificial intelligence, according to him. Naturally, this could be the next step in preparing for a pandemic.
The study’s funding came from the university’s internal AI and big data seed funding program.
Niloofar Yousefi, a postdoctoral research associate in the UCF Complex Adaptive Systems Laboratory, Aida Tayebi, a doctoral student in the UCF Department of Industrial Engineering and Management Systems, Elayaraja Kolanthai, a postdoctoral research associate in the UCF Department of Materials Science and Engineering, and Craig Neal, a postdoctoral research associate in the UCF Department of Materials Science and Engineering, were also co-authors.
Garibay joined UCF’s Department of Industrial Engineering and Management Systems, a division of the College of Engineering and Computer Science, in 2020 after receiving her doctorate in computer science from the university. She had worked in information technology for the UCF Office of Research for 16 years before that.
