Machine learning models can generate and improve the structure of molecules that could be used for therapeutics
To deal with the time-consuming and tedious situations of creating and improving the medications, the researchers at MIT used machine learning to automate the process. The motivation behind this was to replace the inefficient human modification process of designing molecules with automated iteration and assure the validity of the molecules they generate.
Trained machine learning models
The team of researchers trained their machine learning model on 2,50,000 molecular graphs which are basically detailed images of the structure of a molecule. The researchers then had the model generate molecules, find the best base molecules to build and design new molecules with improved properties. They found that the model was able to complete these tasks more effectively than other systems designed to automate the drug design process.
Interestingly, when the models tasked with generating new valid molecules, each one of the models created the valid molecules. Secondly, when the models were told to find the best base molecule which is known as a lead molecule, the models again outperformed other systems.
Lastly, when the model was told to modify 800 molecules to improve them for those properties but keep them similar in structure to the lead molecule. Then around 80 per cent of the time, it created new, similarly structured molecules that scored higher for those two properties than did the original molecules.
Heading ahead, the research team will test the model on other pharmaceutical properties and work to make a model that can function with limited amounts of training data.