 Let me show you an example of such pharmacopore development that colleagues of ours did. These are all dopamine receptor targeting drugs, and they're agonists. There are a bunch of common elements here. First, they all have a so-called tertiary amine group. This is not a chemistry class, so forget about it, but it's a small N in the lower left, very lower right for you there. They all have at least one aromatic ring. The nitrogen is always at the same distance, three bonds away from that aromatic ring. There are polar groups on all the aromatic rings here, and there's a hydrophobic tail on all those nitrogen groups. Again, not all of these properties are necessarily important, but we have 10 molecules that all bound and appear to have a bit of that property. So let's try to distill that down. So then people created a pharmacopore model here where we had the hydrogen bond donors and acceptors. The donor in red, the acceptors in purple. You had the aromatic rings here that we represented in green, and then we had that hydrophobic part that was the tail on the amine, and then the specific positions of say that tertiary amine. If you now take that molecule and then just look at it, it turns out that we can't twist the molecule anywhere, right? It points the molecule with bump into each other. Remember what I said about the flexibility. We don't want molecules that are too flexible because they're going to lose too much free energy when we bind. Sorry, we're going to gain too much free energy when we bind because we're losing entropy. So maybe if I try to fill out those cavities just with large hydrophobic groups or something, again, just to make the molecule less mobile and make sure that it fills out the cavity when I bind. So that would correspond to identifying a number of sites around it and then just try to add some large, say, methyl group, CH3 groups or so around it. The hope here is that eventually they should fill a binding pocket. But there is one important caveat here. This far we don't know what the binding pocket is because I haven't talked to you about the structure of the actual targets yet. This turned out to be pharmacophores that worked reasonably well at least. But typically we don't just do pharmacophore development. We want to start looking at the actual site where we're going to bind things, in particular because there are now, I was about to say, it's so easy to determine membrane proteostructure, it isn't, but it's much easier than it used to be. And if we know the receptor binding site, we can benefit a lot from looking at that site and maybe try to design a key to fit the lock.