![]() This submission contains an implementation of: "Molecular de-novo design through deep reinforcement learning." Journal of cheminformatics 9.1 (2017): 1-14. Prior Network and supervised learning from the GUI.įor more detail, Please see the attached PDF manual. GUI for creating Prior Network and supervised learning: Not only SMILES files but also MDB and SDF can be used as input data to create.Scoring samled compounds: SMILES sampled by the generative model can be converted to 3D and scoring functions can be applied.You can monitor the structure of the generated SMILES and its score. ![]() ![]() Executing and monitoring of reinforcement learning: Various parameters can be set from the GUI to execute reinforcement learning.Scoring functions of MOE: MOE descriptor, QSAR model, fingerprint model, pharmacophore, consensus model, and docking score can be used as scoring.MOE's REINVENT interface provides the following features along with an easy-to-use GUI. The models supervised-learned by the compound library can be further adjusted by reinforcement learning that incorporates scoring functions such asįingerprint similarity and activity prediction models. REINVENT is a SMILES generative model based on the Recurrent Neural Network implemented in the programming language Python.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |