That is sort of the thing about receptors, it is really had to know what will actually work. When you run these types of programs you'll get a number of possibilities that the simulator will spit out, that be from an AI or more traditional stat generated model. There were a few articles very similar to this with Coronavirus cures. They then have to go work in a well designed clinical trial, against a placebo, and then in humans, which is way more unlikely. It's most likely that all 9 don't work, unfortunately. I'm not saying the fact they found them was luck, the luck is if they actually work.