 So, once you have something on that grid that looks well, we're going to need to spend a little bit more time because when we do things on the grid, we ignored the internal rotation of the bonds. So, now I might pick my 1% best targets and for those targets I actually allow the bonds to be flexible and rotate a little bit, spend more time on them so that I can hopefully make better assessments whether they are good binders or not. Eventually at some point I'm going to get a ranked list of compounds. The units of that list could be calories per mole. Don't assume that it's an actual binding energy that should rather be seen as a relative number, something that has a very low binding energy here. Low is good, right? If it's low, it's likely to be better than the other, says the computer. The likelihood that you would get this exact order even of the top 10 compounds if you go into the lab is pretty close to zero. So, once you've decided that you now have something like look promising here, you likely want to test say the top 10 compounds or maybe choose wisely so you test five or six different compounds from the top 20 ones. Based on those, you might see that one of them had a small effect when you tested it in the lab. That's awesome because remember, you only need something. If we see that the one that had slightly more hydrogen bonds on average than the others made better, then we can decide to refocus our docking, right? So let's now focus even more on the molecules that had good hydrogen bonds and possibly add an extra weight to that. Then we make a new round where we test even more molecules and see if we can understand this better and four weeks later we send off a new order to the lab and then have them test 10 new molecules. So each such round is fairly expensive, but it still orders a magnitude cheaper than testing everything experimentally.