There’s a world of chemical compounds out there, some as large as 1060, any of which could potentially function as small-molecule drugs. It’s a monumental task for scientists to assess each one – think finding one particular grain of sand on a beach. But, thanks to the latest advancements in artificial intelligence, the challenge is becoming less daunting. Researchers now have the tools to sift through potential drug candidates far more efficiently.
Enter leading figure in this exciting field, MIT Associate Professor Connor Coley, PhD ’19. Straddling the world of chemical engineering and computer science, Coley wears many hats. He holds joint appointments in these departments and in the MIT Schwarzman College of Computing. In this interesting converging point of science, he works diligently to design and analyze potential chemical compounds. He then creates computational models that predict how these compounds can be created. It’s a general approach, but as Coley explains, “The primary application that we think about is small-molecule drug discovery.”
However, Coley’s road to this intersection of chemistry and AI wasn’t straight or narrow – it was carved over time by years of studies and a constant leaning toward science that can be traced back to his family’s evident influence. His father is a radiologist, his mother studied molecular biophysics and biochemistry before attending the MIT Sloan School of Management, and his grandmother was a math professor. Raised within a family culture of science, it’s no wonder that Coley developed a strong inclination towards it, inspiring him to major in chemical engineering at California Institute of Technology (Caltech) and later, pursue a PhD in the same field at MIT.
At MIT, Coley focused on enhancing automated chemical reactions, teaming up machine learning with cheminformatics, a discipline that incorporates computational methods into chemical data. This allowed him to plan out how new drug molecules could be created. He also designed innovative hardware to automate these reactions. Throughout his studies and research, he garnered experience, especially working on the DARPA-funded Make-It program aimed at refining the synthesis of medicines using machine learning and data science.
After obtaining his PhD at 25, Coley stayed on at MIT, assuming a faculty position – he found the opportunities it offered far too enticing to resist. “MIT is a very special place in terms of the resources and the fluidity across departments,” he notes. Despite advice suggesting moving to a new academic environment, the compelling dynamism of MIT with its student quality, scientific community, and interdisciplinary collaborations was hard to pass up.
Before finally settling into his faculty role, he completed a postdoc at the Broad Institute, adding to his expertise in chemical biology and drug discovery. He worked on identifying small potential drug candidates from billions of DNA-encoded library molecules. Now in his own lab at MIT, Coley and his team harness the power of AI to create or improve existing compounds with therapeutic potential and design new ones with desirable characteristics. They’re focused on pairing challenges in chemistry with computational solutions and have developed several computational models, like ShEPhERD – now used by pharmaceutical companies to discover new potential drugs, and FlowER, another generative AI model designed to predict reaction products. Line by line, these projects are writing a new script for AI in drug discovery.
Under Coley’s wing, students are exploring various facets of optimizing chemical reactions, like computer-aided structural elucidation, laboratory automation, and experimental design. “Through these different research threads, we hope to advance the frontier of AI in chemistry,” says Coley. And so, they continue to rewrite the shockwaves in the ever-evolving field of AI and Chemistry.
If you’re interested in the potential of using AI automation for your business, consider exploring solutions with implementi.ai.
For more, you can read the original news article about Connor Coley’s work here.
This website uses cookies.