Chemists of Princeton University use machine learning to predict the future of chemical reactions

6th March 2018

Chemists of Princeton University use machine learning to predict the future of chemical reactions

A group of researchers led by Abigail Doyle, the A. Barton Hepburn Professor of Chemistry at Princeton University, and Spencer Dreher of Merck Research Laboratories, has found a way to predict reaction yields accurately while varying up to four reaction components by using an application of artificial intelligence known as machine learning. They have turned their method into software that they have made available to other chemists. They published their research Feb. 15 in the journal Science.

“The software that we developed is designed to accommodate any reaction or substrate type,” said Doyle. “The idea was to let someone apply this tool and hopefully build on it with other reactions.”

Vast resources and time are expended to make synthetic molecules, often in a largely ad hoc manner, she said. Using this new software, chemists can identify high-yielding combinations of chemicals and substrates more cheaply and efficiently.

“We hope this will be a valuable tool in expediting the synthesis of new medicines,” said Derek Ahneman, who completed his chemistry Ph.D. in Doyle’s lab in 2017 and works for IBM.

After Ahneman finished his degree, Estrada continued the research. The goal was to create software that was accessible not only to computer experts like Ahneman and Estrada but the broader synthetic chemistry community, said Doyle.

She explained how the software works: “You draw out the structures — the starting materials, catalysts, bases — and the software will figure out shared descriptors between all of them. That’s your input. The outcome is the yields of the reactions. The machine learning matches all those descriptors to the yields, with the goal that you can put in any structure and it will tell you the outcome of the reaction.

The paper, “Predicting reaction performance in C–N cross-coupling using machine learning” by Derek Ahneman, Jesús Estrada, Shishi Lin, Spencer Dreher and Abigail Doyle, was published Feb. 15 in the journal Science. Financial support was provided by Princeton University, an Amgen Young Investigator Award and a Camille-Dreyfus Teacher Scholar Award.

For more information visit

View more Industry News

Latest News