AI dramatically speeds up the chemical process of building molecules that will be tomorrow’s medicines, agrichemicals or materials

Tim Cernak’s lab at the University of Michigan dramatically speeds up the time-consuming chemical processes that build the molecules that will become tomorrow’s drugs, pesticides or materials, with the help of artificial intelligence and massive human hands. I made a discovery.

The discovery, published in the February 3 issue of Science, is the culmination of years of chemosynthetic and data science research by the Cernak Lab in the Department of Pharmacy and Chemistry.

The goal of this research was to identify the key reactions in the synthesis of molecules, ultimately reducing the process to as few steps as possible. Alkaloid synthesis was achieved in just three steps. Previous syntheses required 7-26 steps.

“For example, it’s difficult to create chemical structures with atoms in the right places to provide an effective, non-toxic drug,” says Cernak, assistant professor of medicinal chemistry and chemistry. “We need a chemosynthetic strategy based on chemical building blocks that can actually be purchased and pieced together using chemical reactions.”

This achievement has powerful implications for accelerating drug development.

Cernak likened the structure of these complex molecules to playing chess. To reach the end of the game, you have to coordinate a series of moves. The possible moves are almost endless, but there is a logic you can follow.

“We developed a logic here based on graph theory to get to the end as quickly as possible,” he said.

Cernak and colleagues used the SYNTHIA Retrosynthesis software, which provides scientists with a database of millions of molecular structural pathways, or steps, and formulas. You can now work with synthetic data.

Researchers use the algorithms they develop to curate the data so they can take high-impact steps along the path, or critical steps, and progress towards completing the integration but eventually the whole process. identified an inefficient step.

“I hope this research will lead to better drugs,” said Cernak. “So far, only a limited number of molecular structures have been readily accessible for chemical synthesis.”