 So how do we know that things work? Well, in general, you're going to need to determine a structure of it, just showing that something by and somewhere is not going to be sufficient for your boss to trust you. With COVID, a lot of things that happened just the last few months or the last year at least. So already in mid-2020, there was a structure from Shanghai Tech where they determined a so-called co-crystal. That is a crystal of the protein with its partner molecule bound. In this case, it's a small protease, published in April 2020, I think. Do you see that the small green molecule there, or in this case, the electron density over here? So this was an x-ray structure determined to a resolution of roughly two angstroms, super high resolution. We are ecstatic in this case because you see exactly how the molecule is bound. And at this point, ideally, you now have a process where you understand the biology. You had predicted a binding site where you thought things would interact. If you decide a docking algorithm that was apparently able to target that, you were able to make predictions and you were able to validate those predictions and show that you really had something binding there. You still don't know if it actually has any biological effect though. This just shows that it's bound there, affinity. There are several other groups that have published structures like this. This is just another one, so let me move on to the next step, efficacy. The only way to show efficacy is really to take this in the lab and show that if you're adding these compounds that you thought bound and everything, will they actually reduce versus the viral load if you're having actual cell lines? In this case, it turns out that a few of those molecules actually did. Which again, this is amazing because now we have a molecule where, A, we know that it binds and B, that it's likely to have an effect. The problem is that these are simple cell lines, isolated cells in test tubes basically. Very little says whether this is going to work in a mouse. Why? Well, remember, admethox for instance, right? So there are so many steps left on the way. We're still not sure whether it's going to be a good drug, but things are starting to look promising. The better a binder it is though and the more effective it is, the lighter it is that it will be able to survive those later stages. So could the computer possibly help us here too?