MIT Model Predicts Chemical Reaction’s Turning Point in Under a Second Using Machine Learning

Chemists designing new chemical reactions often seek a crucial piece of information: the reaction’s transition state. This is the point of no return that dictates whether a reaction will proceed. Knowing this allows chemists to optimize conditions for desired reactions. However, traditional methods for predicting this transition state are complex and computationally intensive.

Researchers at MIT have developed a machine-learning model capable of predicting these transition states with high accuracy in under a second. This breakthrough could significantly streamline the design of chemical reactions for creating various useful compounds, including pharmaceuticals and fuels.

“We’d like to ultimately design processes to take abundant natural resources and turn them into molecules that we need, such as materials and therapeutic drugs. Computational chemistry is really important for figuring out how to design more sustainable processes to get us from reactants to products,” says Heather Kulik, the Lammot du Pont Professor of Chemical Engineering and a professor of chemistry at MIT, and the senior author of the new study.

The research, led by former MIT graduate student Chenru Duan, former Georgia Tech graduate student Guan-Horng Liu, and Cornell University graduate student Yuanqi Du, is published in Nature Machine Intelligence.

A Leap in Efficiency

The transition state represents the energy threshold a chemical reaction must overcome to proceed. These states are incredibly brief, making them nearly impossible to observe experimentally. Quantum chemistry techniques can calculate these transition states, but they demand significant computing power and time, often taking hours or even days for a single calculation.

Kulik notes the irony: “Ideally, we’d like to be able to use computational chemistry to design more sustainable processes, but this computation in itself is a huge use of energy and resources in finding these transition states.”

Building on previous work, where a machine-learning strategy was developed to predict transition states, the team has now created React-OT. Unlike its predecessor, which required multiple iterations and a “confidence model,” React-OT uses linear interpolation to estimate the initial transition state, significantly reducing the number of steps needed.

“A linear guess is a good starting point for approximating where that transition state will end up,” Kulik explains. “What the model’s doing is starting from a much better initial guess than just a completely random guess, as in the prior work.”

React-OT can make predictions in approximately 0.4 seconds with about 25% higher accuracy than the previous model, eliminating the need for a confidence model.

Chenru Duan emphasizes the practical implications: “That really makes React-OT a practical model that we can directly integrate to the existing computational workflow in high-throughput screening to generate optimal transition state structures.”

Broad Applicability

The model was trained on a dataset of 9,000 chemical reactions, encompassing reactants, products, and transition states calculated using quantum chemistry methods. It demonstrated strong performance on both familiar and novel reactions, including those with larger reactants containing side chains.

Kulik highlights the importance of this generalization: “This is important because there are a lot of polymerization reactions where you have a big macromolecule, but the reaction is occurring in just one part. Having a model that generalizes across different system sizes means that it can tackle a wide array of chemistry.”

The team is currently expanding the model’s capabilities to predict transition states for reactions involving elements like sulfur, phosphorus, chlorine, silicon, and lithium.

Markus Reiher, a professor of theoretical chemistry at ETH Zurich, who was not involved in the study, commented: “To quickly predict transition state structures is key to all chemical understanding. The new approach presented in the paper could very much accelerate our search and optimization processes, bringing us faster to our final result. As a consequence, also less energy will be consumed in these high-performance computing campaigns. Any progress that accelerates this optimization benefits all sorts of computational chemical research.”

The MIT team has also created an app to allow other scientists to leverage this technology in their own research.

“Whenever you have a reactant and product, you can put them into the model and it will generate the transition state, from which you can estimate the energy barrier of your intended reaction, and see how likely it is to occur,” Duan concludes.