Using GANs with Adaptive Training Data to Search for New Molecules
Share
Training runs with molecules of 20 atoms or less. Results are shown for control (blue) and drug replacement with recombination (green). A) Histogram showing number of new molecules produced in control run for different drug-likeness scores. B) Histogram showing number of new molecules produced in our approach using updates to the training data for different drug-likeness scores. C) A few sample new molecules from the drug replacement with recombination run.
The Science
Generative machine learning models, including GANs (Generative Adversarial Networks), are a powerful tool toward searching chemical space for desired functionalities. Here, we have presented a strategy for promoting search beyond the original training set using incremental updates to the data. Our approach builds on the concepts of selection and recombination common in Genetic Algorithms and can be seen as a step towards automating the typically manual rules for mutation.
The Impact
Our results suggest that updates to the data enable a larger number of compounds to be explored, leading to an increase in high performing candidates compared to a fixed training set. Introducing replacement and recombination into the training process empowers the use of GANs for broader searches in drug discovery.
PI/Facility Lead: Debsindhu Bhowmik Funding: DOE ASCR, ECP Publication for this work: A. E. Blanchard, C. Stanley and D. Bhowmik; Using GANs with adaptive training data to search for new molecules, J Cheminform (2021) 13:14; DOI: https://doi.org/10.1186/s13321-021-00494-3
