AI System Could Become New Chemistry Assistant

A German team of researchers developed an AI capable of dissecting chemical reactions through retrosynthesis.
Shelby Rogers

Artificial intelligence systems have been developed for nearly every other facet of scientific disciplines, but one new AI tool could help chemists out enormously. A team of researchers developed a deep learning computer program that details the sequences of chemical reactions needed to create organic molecules and compounds. 

While this isn't chemistry's first AI breakthrough (and certainly not the first piece of software developed for chemical reactions), the development has already been called "landmark" by scientists not associated with the study.

“What we have seen here is that this kind of artificial intelligence can capture this expert knowledge,” said Pablo Carbonell, who designs synthesis-predicting tools at the University of Manchester, UK, and was not involved in the work.

The software takes its inspiration from AI systems that have surpassed human ability in online gaming. Software like "AlphaGo" combine Monte Carlo Tree Search and deep neural networks that use machine learning and artificial intelligence. Engineers from the University of Muenster in Germany applied that same technology to chemical synthesis (or retrosynthesis) in order to plan reactions. 

Marwin Segler served as the lead author on this study. 

"Retrosynthesis is the ultimate discipline in organic chemistry," Segler explained. "Chemists need years to master it -- just like with chess or Go. In addition to straightforward expertise, you also need a goodly portion of intuition and creativity for it. So far, everyone assumed that computers couldn't keep up without experts programming in tens of thousands of rules by hand. What we have shown is that the machine can, by itself, learn the rules and their applications from the literature available."

And the machine effectively knows roughly 12.4 million single-step organic chemistry reactions. This allows it to predict the chemical reaction from any one step and then applies the same practices to multi-step synthesis. 


"The deep neural networks are used for predicting which reactions are possible with a certain molecule," said Mike Preuss, an information systems specialist and co-author of the study. "Using the Monte Carlo Tree Search, the computer can test whether the reactions predicted really do lead to the target molecule."

The team tested the software in a double-blind trial. They wanted to see whether practicing chemists could differentiate the AI's synthesis from those created by human chemists. Over 45 organic chemists from institutes in both China and Germany participated. The chemists had no preference for either synthesis. 

"The idea is actually about 60 years old." says Segler, "People thought it would be enough, as in the case of chess, to enter a large number of rules into the computer. But that didn't work. Chemistry is very complex and, in contrast to chess or Go, it can't be grasped purely logically using simple rules. Added to this is the fact that the number of publications with new reactions doubles every ten years or so. Neither chemists nor programmers can keep up with that. We need the help of an 'intelligent' computer."

Segler said that the team has already been in discussions with a handful of pharmaceutical companies. However, he assured organic chemists that this technology won't be putting them out of work. Rather, Segler said the AI will serve as the ultimate lab assistant. 

"We hope that, using our method, chemists will not have to try out so much in the lab," Segler adds, "and that as a result, and using fewer resources, they will be able to produce the compounds which make our high standard of living possible."

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board