Of all the possible chemical compounds, it has been estimated that between 1020 and 1060 May have potential as small-molecule drugs.
It would be too time-consuming for chemists to experimentally evaluate each of those compounds. Therefore, in recent years, researchers have begun to use artificial intelligence to help identify compounds that might make good drug candidates.
One of those researchers is MIT Associate Professor Connor Cooley PhD ’19, Class of 1957 Career Development Associate Professor with shared appointments in the departments of Chemical Engineering and Electrical Engineering and Computer Science and the MIT Schwarzman College of Computing. His research straddles the line between chemical engineering and computer science, as he develops and deploys computational models to analyze large numbers of potential chemical compounds, design new compounds, and predict the reaction pathways that generate those compounds.
“This is a very general approach that can be applied to any application of organic molecules, but the primary application we think about is small-molecule drug discovery,” he says.
Intersection of AI and Science
Koli’s interest in science stems from his family. In fact, he says, his family includes more scientists than non-scientists, including his father, a radiologist; His mother, who earned degrees in molecular biophysics and biochemistry before attending the MIT Sloan School of Management; and his grandmother, a mathematics professor.
As a high school student in Dublin, Ohio, Cooley participated in Science Olympiad competitions and graduated from high school at the age of 16. He then went to Caltech, where he chose chemical engineering as a major because it provided a way to combine his interests in science and mathematics.
During his undergraduate years, he also showed an interest in computer science, working in a structural biology laboratory using the FORTRAN programming language to help solve the crystal structure of proteins. After graduating from Caltech, he decided to pursue chemical engineering and came to MIT in 2014 to begin his PhD.
On the advice of Professors Claves Jensen and William Green, Collie worked on ways to optimize automated chemical reactions. His work focused on combining machine learning and cheminformatics – the application of computational methods to analyze chemical data – to plan reaction pathways that could create new drug molecules. He also worked on designing hardware that could be used to perform those reactions automatically.
Part of that work was done through a DARPA-funded program called Make-It, which focused on using machine learning and data science to improve the synthesis of drugs and other useful compounds from simple building blocks.
“That was my real entry point into thinking about chemoinformatics, thinking about machine learning, and thinking about how we can use models to understand how different chemicals can be made and what reactions are possible,” says Coley.
Koli began applying for faculty jobs while still a graduate student and accepted an offer from MIT at the age of 25. He received mixed advice for and against taking a job at the same school where he attended graduate school and ultimately decided that a position at MIT was too attractive to turn down.
“MIT is a very special place in terms of resources and fluidity across departments. MIT seemed to be doing a really good job supporting the intersection of AI and science, and it was a vibrant ecosystem to live in,” he says. “The caliber of students, the enthusiasm of students and the incredible strength of collaboration certainly outweighs any potential concerns about being in one place.”
chemistry intuition
Coley deferred the faculty position for a year to pursue a postdoc at the Broad Institute, where he sought more experience in chemical biology and drug discovery. There, he worked on methods to identify small molecules from billions of candidates in DNA-encoded libraries that can have binding interactions with mutated proteins associated with diseases.
After returning to MIT in 2020, he created his own laboratory group with the mission of deploying AI not only to synthesize existing compounds with therapeutic potential, but also to design new molecules with desirable properties and new ways to make them. Over the past few years, his laboratory has developed a variety of computational approaches to tackle those goals.
“We try to think about how best to pair a challenge in chemistry with a potential computational solution. And often this pairing inspires the development of new methods,” says Coley. A model developed by his laboratory, known as ShEPhERD, was trained to evaluate potential new drug molecules based on the three-dimensional shape of the drug molecules and how they would interact with target proteins. This model is now being used by pharmaceutical companies to help discover new drugs.
“We’re trying to give the generative model more medicinal chemistry intuition, so that the model is aware of the right parameters and ideas,” says Colley.
In another project, Coley’s lab developed a generative AI model called FLOWER, which can be used to predict reaction products that result from combining different chemical inputs.
In designing that model, researchers developed an understanding of fundamental physical principles, such as the law of conservation of mass. They also force the model to consider the feasibility of intermediate steps that need to occur on the way from reactants to products. The researchers found that these constraints improved the accuracy of the model’s predictions.
“Thinking about those intermediate steps, the mechanisms involved, and how the reaction evolves, is something that chemists do very naturally. It’s how chemistry is taught, but it’s not something that models think about naturally,” says Cooley. “We have spent a lot of time thinking about how to ensure that our machine-learning models are based on an understanding of reaction mechanisms, in the same way as an expert chemist would.”
Students in his laboratory also work on many different areas related to the optimization of chemical reactions, including computer-aided structure elucidation, laboratory automation, and optimal experimental design.
“Through these many different research threads, we hope to push the limits of AI in chemistry,” says Coley.
