Researchers at the University of North Carolina at Chapel Hill Eshelman School of Pharmacy have demonstrated an artificial-intelligence design that can teach itself how to design new drug molecules and has the potential to dramatically accelerate the design of new drugs.
Called “Reinforcement Learning for Structural Evolution” (ReLeaSE), the AI is in the form of an algorithm which has been configured to work with a computer program, based on two neural networks which function like a teacher and a student. The teacher network understands the syntax and linguistic rules required to decipher the chemical structures for about 1.7 million known biologically active molecules, while the student network learns and becomes better at proposing new molecules that could form the basis of new medicines.
“If we compare this process to learning a language, then after the student learns the molecular alphabet and the rules of the language, they can create new ‘words,’ or molecules. If the new molecule is realistic and has the desired effect, the teacher approves. If not, the teacher disapproves, forcing the student to avoid bad molecules and create good ones,” said lead researcher Alexander Tropsha.
Drug researchers currently use virtual screening to evaluate existing large chemical libraries, but the method only works for known chemicals. However, ReLeASE has the unique ability to create and evaluate new molecules.
“A scientist using virtual screening is like a customer ordering in a restaurant. What can be ordered is usually limited by the menu,” said researcher Olexandr Isayev. “We want to give scientists a grocery store and a personal chef who can create any dish they want.”
The research paper, “Deep reinforcement learning for de novo drug design,” was published in the journal Science Advances.