 So in general simulations have the advantage of showing us some kinetics. But in particular for the free energy calculation we actually gave up that, right? I only get the binding energy off the ligand in the binding site and I put it in the binding site from the start. Ideally in a simulation I would like to see the entire movie that is putting the ligand somewhere in the box and letting it diffuse and find the binding site itself because then I'm not thinking any assumptions of where it's binding. That has been possible in a handful of cases but this is very much state of the art cutting edge research. I'll show you one from the David Shaw group. This is about 10 years ago they took a tyrosine kinase receptor. Remember that I told them but this is just the tyrosine kinase sitting on the inside of the membrane and then we're putting this small drug here in orange and then we hit the play button and see what happens. This takes several microseconds now it's a very long trajectory. Do you see how this molecule is searching testing different conformations? We're not even seeing all the water here but the protein 2 is moving and then eventually it settles in in the right bound state just as if it was a protein folding and finally being happy in its binding pocket. And you know what? This is more similar to protein folding than you think. Just as in protein folding it appears to be an entropic process, right? It's searching and searching and searching. In this case we know the answer. Why? Because we can determine an experimental structure with this ligand bound and it turns out that the end state of the simulation overlaps virtually perfectly with the simulation. It's black and orange in these plots. But what we can do from the simulation now is that we can study for instance the energy. As it's going there what's happening to the energy? Are we gradually going downhill? And it turns out we are. Do you see the parallel between that and protein folding? And protein folding we also had the energy always going down but the reason for the barrier was that it was an entropic search process. This too is an entropic search process. So while we spend a large part of this class studying protein folding it's because it's an ideal simple problem. But the general properties of having molecules testing different conformations and everything that's equally true for drug design and most other conformational transitions we're going to be speaking about and that you will hopefully work on in your future career. What can we get from this? Well this particular means that we're getting kinetics and now we can start to see what's happening when we're crossing that barrier. What is really the rate limiting step that means that it takes some time for us to bind? Is it that we have to squeeze a few waters out of the binding site or is it that I have a particular number of rotating bonds in my molecule and I simply need to search all through all those bonds before it can bind. It turns out that both those are very common properties but in particular getting rid of this water that exists in the pore before I can squeeze out the water and replace them with my small molecule. I'm going to show you another example of this that Ron Drawer did a few years ago. So here they simulated a process where they tried three beta blockers and one agronist against a GPCR and here too we know what the binding site is and then they simulated this for a few tens of microseconds, lots of simulations. Here too they got beautiful results where the experiment overlaps perfectly with the end state of the simulation. You can hardly, it's gray and purple here, they overlap. You can see the difference. But the question is what's happening on the way there? Well it turns out that there were barriers. It took quite a long time for the molecules to find the right state. Even more surprising there were two barriers actually. And the first barrier here was the one that took longest time and it was fairly, it wasn't all the way down in the pocket. It was fairly far out in the pocket. So this small molecule was searching and rotating and was screwing around about, well one or two nanometers away from the actual state. What then happened in the first barrier is that the molecule dehydrates. So you're getting rid of all the water in the molecule and then we're able to get slightly further in. But that is still not the bound state. Remember, we know what the bound state should be from the x-ray and the structure in this case. And what they found after a few tens of nanoseconds is that eventually the molecule passes the second barrier. The origin of that second barrier is not the hydration, but you see these two green dots on the side here. In this particular case there are kind of two amino acids sticking in here from the side and these are closing off the binding site and they need to move away opening in a path to the binding site, which is just entropy. We have to wait a long time for that to happen and then it closes on the binding site again. It's a beautiful example of something you can see in a simulation but we would never be able to see that just with free energy calculation. The free energy calculation might tell me that it's really good to be bound here, but the free energy calculation can never tell me how long it will actually take for this molecule to go in and out. But just to stress, this is not quite science fiction for now, but these type of studies are far too costly to apply on any type of high throughput scale. Don't start with simulations like this if anybody is asking you to study protein design or drug design.