Computational docking (figuring out what drug molecules bind strongly to) is still not solved. In order to find drugs that 1. work well and 2. aren't toxic we n

Computational docking (figuring out what drug molecules bind strongly to) is still not solved. In order to find drugs that 1. work well and 2. aren't toxic we need robust docking estimators that work well over a wide variety of proteins. My understanding of the space (which comes from undergrad comp bio, so please someone feel free to correct me) is that just having a molecule that binds well to your target doesn't mean it's going to work well as a drug, because there are delivery and toxicity concerns.