Chemical predictions can help prevent unwanted reactions or improve efficacy of industrial processes, among other applications, but predicting how chemicals interact can be costly and time-consuming — especially as how chemicals may react in one environment may behave differently in another. A novel experimental modeling approach may help simplify the process by not only increasing the number of chemical predictions, but also the transferability of such foresight.
A study describing the work was published on Sept. 14 in JACS Au .
"The ability to precisely control reaction rates and selectivity is of fundamental importance in organic chemistry," said corresponding author Hiroaki Gotoh, associate professor, Faculty of Engineering, YOKOHAMA National University. "Our ultimate goal is to predict where and from which side drug candidates, natural products and functional molecules will react before synthesis. This will help reduce trial and error, and accelerate the discovery of new molecules and synthetic routes."
He explained that such control enables better design of chemicals and understanding of how chemicals behave when exposed to one another or in different environments. Currently, reactivity is typically measured by the product yields after multistep synthesis processes, which frequently involves purification and competing side reactions that impact the product yield. This makes product yield an imprecise measurement technique — especially as product yield differs across the same molecules used in different synthesis scenarios.
"Moreover, yield-based evaluations often require optimized conditions and the numerous parameters that influence reaction yields are frequently difficult to identify and quantify," Gotoh said. "Consequently, there is a pressing requirement to establish methods for measuring reaction rates in complex, multistep reaction systems and for analyzing their correlations with molecular structure."
In this study, the team focused on complex molecules that can contain multiple ketone groups. These are double-bonded oxygen and carbon atoms that serve as functional units that react to reduce the complex molecule and create new molecules. Product yield measurements do not provide information about which ketone groups reacted, nor from which side of the group.
"We wanted to predict both where a ketone reduction would occur and which product would form, while overcoming the fact that conventional yield data do not cleanly measure a molecule's inherent activity," said first author Daimon Sakaguchi, a doctoral student in the Gotoh laboratory.
The researchers turned to competition experiments, which are simple designs that can reveal quite a lot of information, according to Sakaguchi. By mixing the complex molecules in varying reagent amounts of sodium borohydride in methanol, they could measure which ketone group reacted first — allowing the researchers to develop a kinetic scale, or reaction-rate map, of the ketone groups. Combining this with three-dimensional electronic information measured from around each reacting site, the researchers developed a model to predict chemical reaction rates. They trained it only on simple molecules, but it correctly predicted both the main reaction site and the side from which the reaction occurred in all six more complex molecules containing two ketone groups.
"We found that simple competition experiments can produce reaction-rate data that connect laboratory measurements, quantum-chemical calculations and machine learning," Sakaguchi said. "Our reported curated kinetic database and descriptor workflow provide a foundation on which research can perform further analyzes, construct new predictive models and test mechanistic hypotheses for complex reduction models."
Collaborators on the paper include Taisei Kawasaki, Mayu Itakura and Chihiro Tada, all of whom are affiliated with the Department of Chemistry and Life Science, YOKOHAMA National University.
The Tokyo Ohka Foundation for the Promotion of Science and Technology, the Next Generation AI Human Resource Development Project, the Japan Science and Technology Agency and the Japan Society for the Promotion of Science funded this research.
