A policy gradient approach for discovering novel drug candiates with desirable QSAR properties. Specifically we use the model to explore around promising drug canidates for recapitulating native function in p53 y220-c. In this setting we have a set of sucessful drugs which we would like to expand/iterate on in a ML-guided manner.
If anyone is coming from the thesis please look at ThesisCode, everything is organizationally reworks and ThesisCode has all thats important