Massachusetts Institute of Technology (MIT) researchers are using machine learning to automate the process of developing and improving drugs.
The team trained a machine learning model on 250,000 molecular graphs, detailed images of a molecule’s structure. The model generated molecules, found the best base molecules from which to build, and designed new molecules with better properties.
The researchers found the model was able to complete these tasks more effectively than other systems designed to automate the drug design process.
In addition, the model was asked to modify 800 molecules to improve them for certain properties while keeping them similar in structure to the lead molecule. About 80% of the time, the system created new, similarly structured molecules that scored higher for those properties than did the original molecules.
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA