Inspired in AutoGrow4, the framework uses in silico RDKit reactions to mutate compounds. A multi-objective evolutionary algorithm explores these mutations to optimize the compounds. Besides generic optimization goals, such as the logP, OptSmiles also includes a bitter classifier.