 Yeah, I'm happy to start whenever you guys. Let me just wait one more minute, we're at five. Yes, okay. Then you can start, so it's easy to keep counting. We just had a thunderstorm actually, so I had to move from the room I was in, because I could hear, you could hear the rain. Okay. You move to a different room. We are streaming now. Okay. Okay. No worries, but just to know it. You can start Alex. Yeah, I think so. Today is Thursday afternoon, people start to be tired. Okay. Let's start with the session. Hello to everybody. For me, it's a pleasure to be here. I'm going, I'm Alex Rodriguez for those that miss the first day. And I'm carrying today. And we have in the first talk of this afternoon, we have Professor Sima Khalid. She's from Southampton, right? Yeah. And her research is focused on multi-stage modeling and simulations of bacteria and membrane on cell walls, if I'm not wrong. So she's going to start the talk for the questions. Probably the best solution is that you write it in the chat. And I read them at the end to Sima. And that's all. Please start. Can you start sharing the screen? Okay, can you see it? Yes. Okay, great. Well, thank you very much. I'm going to start by thanking the organizers for the invitation. Of course, it's a, it's quite gutting not to be able to visit Trieste in in person, but it's nevertheless very nice to be to be able to be with you remotely. So I'm going to talk to you about working my group, which is focused on trying to understand bacterial cell envelopes. But let's, let's just have a quick biology lesson before we start. So actually, is the sound okay? I can hear a slight echo. Is it okay? I listen to you perfectly. Okay, thank you. Okay, so firstly, I guess we should ask, why do we want to study bacterial cell envelopes? The answer to that question is for me anyway, it's a question of endless fascination that bacteria which are single celled organisms are still causing us so many problems. We are sophisticated organisms composed of many, many millions of cells. And yet these bacteria, they, they still develop resistance to antibiotics. They are still many pathogenic bacteria and diseases that they cause for which we really have no answer. So that's one cell. So it's very interesting for us to understand how do they do this and how do they protect themselves? Surrounding their cells is what we call the cell envelope. There's two types of bacteria, gram negative and gram positive. I'm going to focus exclusively on gram positive, gram negative, sorry here. Gram negative have a very sophisticated cell envelope. Okay, so this goes around all the cell. They have two membranes. In between the two membranes, they have an aqueous region, which is called the periplasm. Inside this aqueous region is as you can see in yellow in the diagram is the peptidoglycan, which is also called the cell wall. And this is made of a sugar peptide repeating polymer. Okay, and it can be one layer, can be two layers. It can be a few layers, but in, in gram negative, it's quite thin. In the coli, it's between one and three layers thick. The two membranes are not the same. The one on the outside. So the one on the outside is the one that any antibiotic or any molecule which is coming from the outside. It's the first barrier which it encounters is, is quite, it's quite complex on the very outside. So the outer leaflet, it has a big and quite some horrible molecule, which I'll discuss in more detail later called lipo polysaccharide. On the inner leaflet, it has phospholipids. And these phospholipids can be, they can vary in the head group and the tail, but you get the idea. They have polar head groups and they have hydrophobic tails. The tails can be different lengths. They can have double bonds without double bonds. The head group can be anionic or it can be zwitteranek. The inner membrane in contrast has phospholipids on both sides. Okay, so it's a little bit more, a little bit simpler if you like. Okay, so let's, before we go on to discussing why we want to do simulations to understand these, I want to ask sort of address a very, I think a very important question is, which is, why do we want to do simulations? And I think the answer to this question has evolved over time. The reason for doing simulations has changed. So for example, one fundamental that remains is that we want to rationalize and understand at the molecular level some specific experimental observation. So your friend, your colleagues in the lab, they got some result, but they can't understand at the molecular level why they got that result. Right, so we might want to do a simulation of that small part, the molecules involved to understand at that level. Or of course, we might want to study systems under conditions that the experimental guys are having difficulty achieving. Maybe the temperature is very high or there's some, you know, there's some pressure they can't reach or something that's very expensive to do so we can do some simulation. But I think what's really important is as we go forward, we also surely these the first two are very experimental lead, but surely the job of the simulation or the computational chemist, physicist, biologist is not just to be answering to the experimental guys the whole time. We want to develop and put forward new hypotheses. And now we have lots of computing power. We have lots more experimental information, which is coming through that is tangible for us. I think we can start doing that. We can have a hypothesis that we go to the experimental guys and say you test this now. It turns it on its head a little bit. And so I think, you know, curiosity driven science leads us to this. We get to a point where actually it's okay to do simulations of a system to find something we didn't really know existed. We're just looking. After all, many of us are probably scientists. We're the kind of people who picked up the stone when we were little to see what was underneath, right? And I think it's that kind of thing with the calculations that we're doing. We might not always know what we're looking for. And I think there's a place for that. And I'll come back to that right towards the end of my talk. Okay, so back to the outer membrane. So when I started my, my own group at Southampton, many years ago now, one thing we realized very quickly was that models of the bacterial outer membrane always had phospholipids on both sides. And remember, I told you this is not true. On the outside, we have this molecule called lipopolysaccharide. Lipopolysaccharide has four to six hydrocarbon tails connected by two sugars. Some of the, one of these sugars is phosphorylated or that varies depending on the bacterial species. Then we have many, many layers of sugars on top. So the pink and the green and the orange in the diagram is showing you many, many sugars. So around the same time, actually, a bunch of groups, Wampillium, Teresa Suarez, Jeff Clouder and myself, we independently developed models for this outer membrane using actually all three just by pure coincidence. So Teresa was using amber, Wampill was using charm. I was using Gromos and we came to these models and they all agree. That's a great thing. We all have these now atomistic models and actually we're all friends and we work together, which actually is, I think, another very nice aspect of the bacterial simulation community. But I'm not going to go into masses of detail about this model just simply to say that it really does work. And I'll show you that in just one picture here, which is a study we did with Jeremy Lakers Group in Newcastle. Jeremy did experiments where he made membranes with these lipopolysaccharide molecules. They're cross-linked with divalent cations. He treated them with chelating agents. They ripped out the divalent cations and then they put in sodiums. They did not get a stable membrane, right? And when we did the simulations, you can see here, the same thing happened. We ripped out the divalent and we put in sodiums to keep the ionic strength the same, but the membrane broke. You can see water is getting in. So just to sort of tell you that the model is realistic and it's stable. And it's not just my model that does this. One pill's model does it. Theresa's model does it. So between us, we have some very nice models now of these membranes. But okay, that's okay. Fine. But what are we going to do with them, right? That's more important. So one of the first things we started to do is to have a look at this small protein. And the reason for this was when I was a postdoc in Mark's Ansom's group at Oxford, I'd already done lots of simulations of this protein. It was well known to me. And now I wanted to look at it again in these new membranes we had. And at around the same time, work from our collaborator again at Oxford, Carol Robinson, had shown that this protein exists as a dimer, right? Or not. Or sorry, I should say it can exist as a dimer, the full length of the protein. So far, we had only studied this beta barrel that you can see in gray. This sits in the membrane, but they studied the full length of the protein, which has these extra regions, which you can see here. Now the really cool thing about this protein, in my opinion, it's cool is that this region here in the gray box is, this sits in the periplasm. If you recall, I told you the periplasm is the aqueous region, which is in between the two membranes. Not only that, but it's non-covalently bound to the cell wall. And there's an x-ray, there's a crystal structure of this soluble domain of the protein bound to a small portion of the cell wall. And if you recall, I said the cell wall is a polymer of sugars and peptides. So this was nice, right? We have a place we can start and we can build the model and we can bind it to the cell wall. So we built, what we did is we built one complete unit of the cell walls. There were some bits missing. These bits here were missing in the crystal structure. So we built them in. So we have one monomer of the cell wall. And we ran some molecular dynamic simulations and we found it's very stable binding. There were two electrostatic interactions, one involving an arginine and involving an aspartic acid, which were present in the crystal structure. And I'm just showing you here in red and blue, the distance between the protein and those particular residues of the protein. And the cell wall, you can see that it comes off a little bit, but remains really stable. We get that it's maintained. So we're seeing nice stable binding of the protein in the membrane. What we did then is we extended the cell wall a little bit to make a strand. So instead of one monomer, we now have just one sort of a linear polymer. But we found something slightly odd at this point. What we found was that when we run the simulation, we start off with this linker extended and the cell wall is very far from the membrane. But as we run the simulation, the linker contracts, the soluble domain of the protein is lifting up. And actually the cell wall is now interacting with the lower part of the inner leaflet of the membrane. And you know, this is not physiologically what we expect to happen. We know from tomography data that we know we have a good estimate of what the separation is between the membrane and the cell wall. So this was sort of puzzling. But then when we ran simulations of the dimeric version of the protein, so remember the mass spec studies had shown it exists as a dimer, we saw in this case that the cell wall cannot interact with the membrane because the dimeric interface holds it together. It's unable to move up. And so we were getting this kind of scenario here. We were seeing that actually the, in the monomer, we're seeing severe distortion of the cell wall, these single strands of the cell wall that we had. But when we have the dimer, we don't see the distortion. And so at the time, this was the model of the cell wall we had. This is the system we had. And so we kind of thought, well, could it be that binding of the cell wall by the monomer causes distortions, binding of the, or it could mean that when it's in the monomer, it doesn't bind the cell wall. It has to be in the dimer and then it binds the cell wall. But then we also thought, could it be that the reason we're seeing this distortion is because we are using just a small strand of the cell wall. Our model is not biologically realistic. And I think that's something we always have to keep in the back of our mind when we're studying biology is that it's big and it's complicated and we may be missing something. So we went back and made a much bigger model of the cell wall now and we made the model such that we bound the cell wall to itself across periodic boundaries. So we have essentially an infinite and two-dimension cell wall, which is kind of a good mimic of what you would see because it covers the entire cell, right? It doesn't have an end and a beginning as such. And we ran our simulations. We placed the cell wall at a 90 angstrom distance. And we based this on tomography data. We estimated using that. And actually when we ran the simulation, we got no interaction between the cell wall and the protein. So that was, as you might imagine, somewhat discouraging given we thought we made a better model. But then we thought about it and thought, actually, what is missing still? So it turns out that we're missing a protein called, in this here I've called it BLP. It's called Braun's lipo protein. This is the most abundant protein in E. coli. So there's more of this protein than any other. The protein is covalently bound to the cell wall on one side. On the other side, it's anchored via a lipid tail into the membrane. Now, so it looks like this, here it's covalently bound. And it has lipids on the other side, which are embedded in the membrane. Now, when we ran the simulation, the Braun's lipo protein was able, it kinked a little bit, it tilted a little bit in the membrane. It caused the distance between the cell wall and the membrane to reduce a little bit. The soluble part of the protein electrostatically sensed the cell wall. The linker could expand it. And the soluble domain landed on the cell wall. And it found the same electrostatic interactions that we saw in the crystal structure. This happened spontaneously. And again, this was a lesson to us that we missed out initially a very important part of the system. There is lots of this protein. Although interestingly, I have to say, when we ran the system with the dimeric version of the protein, we didn't need Braun's lipo protein every time. Sometimes by itself, the linker region would expand and the soluble domain would interact with the cell wall. And I think that's probably just because we have twice as many electrostatic interactions now. Maybe there's some threshold, right? I mean, it could be. I guess one would feel, you know, you feel sort of tempted to hypothesize maybe that in the scenario where Braun's lipo protein becomes depleted, we know on pay this protein is upregulated. So maybe this is when they form dimers and these dimers are having this effect. But again, I want to be cautious from making these statements now, given that we know we keep, you know, maybe we're missing something else, but anyway, it's something to bear in mind. Okay. So, so far, I gave you one half of the story, just the outer membrane and the cell wall. We know there's another membrane on the other side. And so we decided to make a model of this one too. We took this model protein, TOL-R, which was made by Aaron Cutts, working with Phil Stansfeld and Colin Clantus and Oxford. So this protein is known to bind in its open form to the cell wall, but there is no structure in the open form. This is the close form of this protein. There is an X-ray structure of this one here is a model that these guys made of the open form. So we asked them for the model and we put it into our system. And these were, as far as we know, the first ever simulations reported in which it really, we have a small slice of the bacterial cell envelope. We have both membranes and cell wall and we have proteins from both sides. And we got a very nicely stable system. You can see here, I'm just plotting the center of mass of the on-pay protein, the cell wall and this TOL-R protein. You can see they're very stable. They don't move very much in multiple. You have three independent runs here. And we were able to identify key protein residues that are important for the binding of TOL-R. And encouragingly, these are the same residues that the experimental guys, Colin Clantus's group had identified. However, what they didn't know was how does it bind? So we ran our simulations by placing this protein just below, just below the cell wall. And as we ran the simulations, it was kind of cool actually. We saw these two term and I reach up, kind of grab hold of the cell wall and then pull this bit up a little bit. So the protein formed the full complement of interactions with the cell wall. So we helped what was known by the microbiologists by providing them with some optimistic insights into this is how it does the binding. It reaches up and it attaches. And, you know, we, as I said, we picked out the same residues as being important. So that was all nice. And I'm just going to skip through this slide very quickly. All I want to say here is just to say that we also found that the cell wall, if you look at the bottom picture remains very stable, very flat, sorry, when you have a protein binding from both sides. If you only have a protein binding from one side or the other side, there are undulations in the cell wall. So we think maybe in real life, it's a balance when you get a slight imbalance between, because these proteins are always moving around, right? They're always moving around in the membranes. When you get a slight imbalance, maybe this leads to sometimes in the tomography data, they see these undulations. Maybe it's that that causes it. What I want to finish off with now is just another, just to explain that actually something is still missing in our models. We have the right membranes. We have the right cell wall model, but we didn't get the crowding right so far. We're just looking at one protein in that membrane, one protein in this membrane. There's loads of them. The periplasm is considered to be a gel-like, although that is debated, whether it's quite gel-like or not, but it's certainly not dilute, right? There are many, many proteins in that region. So we're now doing this work, which is very new work. It's not published yet. It's still under review. It's available on BioArchive, where we've made models where in the periplasm, we're putting many proteins. And then we're throwing in this antibiotic called polymix in B1. And this is, I guess, what I was trying to tell you about. Let's put it all together and see what happens. This is that kind of approach. And not only that, but actually this, my post-doc, Kamrado Pedibos, pointed this out that actually it's not just proteins that are in there. There's lots of small molecules, the byproducts of metabolism. There's all sorts of things, little sugars, which help osmoregulation of the cell. So we put all of those in as well, and we tried to get the correct crowding that you would see in real biology. And we saw something interesting, but before I'm going to go into the details of what we saw that was interesting, I'm just going to explain just so it becomes clear. We put this protein called LOL-A into our system. The job of this protein, LOL-A, is to carry Braun's lipoprotein. So I just mentioned Braun's lipoprotein, right? The one that I said is the most abundant one. It has a lipid on one side. And what it does is the lipid part of Braun's lipoprotein enters a hydrophobic cavity of this protein, LOL-A. It carries it through the cell wall and delivers it to the second protein, LOL-B. Okay, we know this. This is known. And earlier in 2018 in my group, we did some simulations where we saw the spontaneous binding of Braun's lipoprotein into the hydrophobic cavity of LOL-A. So in the yellow spheres, I'm showing you the lipid part of the protein. You can see it's going inside the hydrophobic cavity. But here's a crucial bit. We also know that the cavity is fairly indiscriminate. These small hydrophobic molecules can also go in unbind. This is known experimentally and from our simulations. Okay, so bear that in mind. Okay, so then we ran our super-duper simulations with everything in. And I'm just giving you a little gratuitous movie, really, just to show how nice it looks, I guess. We moved the small molecules out, which you can see here in green. This is the polymixin. The antibiotics are in here. And we've got lots of different proteins. Whoops, let's move that on. Now, the interesting thing is the antibiotic polymixin, it's also a lipopeptide. It has a hydrophobic region, tail. We found in our simulations, it's binding in the hydrophobic cavity of LOL-A. Right? And in this diagram here, just showing you the residues, which are of the cavity. And these are indeed the ones that make the most contact with polymixin. You can see it reaches right inside that cavity. And the amino acids, which are making the most contact with polymixin, are the same ones, which are the ones that are making contact with the bronze lipoprotein tail. Right? So because we made a big crowded system, we generated a new hypothesis. Could this be one way in which polymixin can move across the periplasm? We're not saying this is the only way. But could it be that if you have a protein, which is floating around with a big hydrophobic cavity, and you have an antibiotic with a hydrophobic tail, as a chemist, it's intuitive that it will go in that cavity at some point. And maybe it's hijacked. And so we call this the hitchhiking mechanism. Maybe it's just hitching a ride on this protein. As I said, it's probably not always doing this. Maybe it does it every now and then. But we could only see this because we made these complex models. And so with that, I want to conclude to say that the specific conclusions from our work are that bronze lipoprotein plays a key role in maintaining the distance between the cell wall and the outer membrane. I don't want to say we are the only ones who are saying this. This is also known from experimental studies. But the bit that I think is new from us is that that also plays a role in facilitating the non-covalent interactions between other proteins and the cell wall. Like we saw with Ampe, it needed the bronze lipoprotein to tilt and kink and do its business before it could come and interact. We know that toll R forms a stable non-covalent binding interaction with the cell wall. We think polymixin might be transported through the periplasm by hitchhiking on the carrier protein LOLA. It'd be great if some experimental person could test this for us. I'm not scared of being proven wrong. I think that's science, right? We throw ourselves out there and someone proves us right or they prove us wrong. But the important thing is we throw ourselves out there. I'm pretty happy to do that here. We really, one thing that's come out of all of this, and some of this we've learned the hard way. And if you guys can learn lessons from me and don't make the same mistakes we do, then biological details of complexity and crowding are important. And we do have the computer power now and the understanding that we don't have to neglect them. So maybe we should start including these details. And finally, we're working towards a virtual bacterial cell envelope model. We'd like to have an atomistic model and a coarse grain model. I haven't talked about any coarse grain work here. We're doing that too. So we can just as as Villa said in his talk this morning, we can use the microscope to look in and see what's going on. Okay. And with that, I'm going to finish with some acknowledgments. All members of my group. One nice thing about everyone having a big project like developing bacterial cell envelope models is that every member of my group is involved in every project. Everyone sort of, so, you know, there's too many to sort of name each one as we go along, but every single one has been involved and I'm grateful to them all. Grateful to collaborators, experimental and computational various sources of funding and computer time. And I wanted to leave you all with a picture because in this modern COVID world, we can't travel so much. And in case you were wondering how people come to work and what they look like in the UK, well, this is us. Okay. Thank you very much. Okay. Thank you very much. Thank you very much for your nice talk. I think we can start with some of them that are starting to write in the chat. So let me read. Are you going to read those or shall I? You want to, okay. If you want to read it yourself, it's okay if you see that. I don't know why my computer is kind of keeping the chat constantly. Okay. So if we like to perform the look at to the interaction between and bacteria and molecules with equal number. Outer. The back to the back solution or we should consider more detail on the bacteria membrane model near the aqueous drug solution. Sorry, I'm just trying to read the question. Can we consider outer LPS part as the, I don't know. I'm not sorry. I mean, can the person ask the question? I'm muted and ask. Yeah. I think mostly because you know, if it's the outer part, you need to call the women for the interaction. So I think. I think it's really important that we have the details of. I think it's important that we do have the details because if you think of LPS, it's very, very. There's lots of charges, right? The electrostatics are really important and also it has many hydrogen bonding capabilities. So I think if you're going to study any antibiotic coming into the membrane or sort of entering the periplasm, then that's really important to get a handle on. But I should also say it's not very easy because LPS diffuses 10 times slower than phospholipids. It's like, and in real, it's like concrete, right? It doesn't move. So one problem we're having, and I'm discussing this with one pill and others at the moment is we need to find enhanced sampling methods. We need to do something to speed up this movement of LPS without damaging the kinetics of the rest of the system. So I think it's important, but I think also you have to be careful that you have enough sampling. Another question from Manuel Cagone. That's the models have been applied for research new antibiotics. Sorry, can you, can you say that again? I lost it just for a second. Okay. That's your models applied for research new antibiotics? No. No, we're, okay. So we're very new with the antibiotics. The last 10 years in my group has been spent just making the, you know, the sort of in vivo biology parts right. The antibiotic is new. So that's very new work we're beginning to do now. Another question from Sharon T. It's just how large are your atomistic models with multiple protein copies? Okay. Yeah. So they're not, I mean, we're going up to 700,000 atoms. They're not huge. We are building models now which have both membranes, all the crowding and everything. And those are going to be much bigger. We're still in the process of putting those together. Then. So, can we use the bacterial cell of involved model for human systems also? No, I'm afraid not. So the LPS is the LPS outer layer is very specific to bacteria. If you want to look at human membranes, then, you know, they have ganglion sides, cholesterol, finger lipids, all these other lipids which are not present in bacteria. So it needs specific models. Another question. This one from Gallo Ezequiel. Did you evaluate the influence of the lipid inner membrane composition? And did you use a standard mix of phospholipids? It's a very good question. We used a mix of PE, PG and cardiolipid. So we used a mix of lipids which is specific to E coli. But we used it symmetrically. We had the same mix in both leaflets. But a paper came out this year, actually, which is now saying that it's not symmetric actually. So now we have to go back and do that again. Okay, Sima, I have to ask you to read yourself the questions because it looks like I'm having problems with my phone and generating some mess to everybody. So I will just silence myself, me and myself, and let you replay the questions because sorry, I cannot do anything. I don't know what to do. Okay. So a question is asking if the protein can be used as a dimer. So we know that. Is there a crystal structure of the dimer? The answer to that is no, we don't have a crystal structure of the dimer. It was seen in mass spec experiments. And we already had a model of the full-length protein. And then that was used by Carol's group to make a model of the dimer. And then we took the model back from them. So it is actually a model of the whole thing. We're saying you don't need another protein. Is the tagging protein needed? Is the tagging protein needed? By tagging protein, do you mean, you mean bronze wiper protein? Yeah. So there was another protein which was needed. When you were using monomeric form to interact with the lower member. Correct. So, but when you were taking dimer of dimeric form of the protein, that is not always needed that other protein which was tagging. So my question is like in actual cell, whether it is needed or not, because we don't know whether it exists as dimer or monomer. Right. Right. So, okay. So bronze wiper protein. There's lots of it. It's ubiquitous. There's monomeric. Bronze wiper protein. There's lots of it. It's ubiquitous. There's more of this protein than any other. So I think, and the cell does need it to function correctly. What happens is if you have bronze wiper protein mutants, the cell is still viable. It exists. But it's basically it's sick. What happens is more on pay is expressed, but also how do I explain it? So it forms blebs. So the cell starts throwing out little parts of itself. It starts forming these vesicles. So actually it's not so stable. The cell envelope is not happy. Right. So I think it really does need, it does need to be perfectly happy. And for normal growth, it needs this bronze wiper protein. Okay. So when the modeling of dimeric form, you saw that it's not always needed when you take dimeric form. That means what, like when it is in dimeric form, the cell memory is not very happy. No, I think, I think what we're saying is that we, up until now, there has been no explanation for why there would be a physiological reason for a dimer. Okay. There's some controversy in the literature of does on pay exist as a monomer? Does it exist as higher oligomers? Right. This controversy has been going on forever since I got into this field as even as a postdoc many years ago. So I think one plausible, one, one sort of plausible explanation is that it's largely a monomer, which is interacting with the cell, with the cell wall. But when bronze lipoprotein is depleted for whatever reason, or it moves away, maybe, then you can see a reason for these on pays to come together and form a dimer because it cannot form a covalent bond with the cell wall. But when two of them come together, they form a much stronger non covalent interaction. Yes, God, you know, that's just conjecture from my part, right? I mean, I don't know for sure. Yeah. Okay. Thank you. Thank you. Okay. What is the lifetime of the cavity involved in the hitchhiking mechanism? So we, we see this interaction is put stable over sort of, you know, these are not long simulations. So over 200 nanoseconds or so we see it's inside. It doesn't come. Once it goes inside in our simulations, the lipoprotein into the cavity, it's happy. It never comes back out. It could be that when we excuse me, we're extending the simulations. It could well be that when we run the simulations longer, it will come back out. But so far we haven't seen this. Excuse me. And last one is asymmetrical membrane have a lot of physical problems. So it's better to use symmetrical compositions. If the question is not focused on the inner membrane. Okay. I think that's just that that's a statement rather than a question. Yeah. I would agree with that probably if you're not focused exactly on the interactions, it makes sense to use a symmetrical composition. I think how do you choose the ratio of different lipid components in the membrane from lipid omics studies. This data is known from experiments. So you take the experimental studies. Are you reporting the PDB atomistic structure file of E coli membrane. That's a good question. No, we don't have the membrane, but some of our trajectories are available on some web sites, which is in the sporting information. And also if anybody wants those atomistic models, they can email me and they can have them, of course. Okay. I think, I think I've gone through all of them now. Yeah. I think so. Sorry for the problems with my audio. I think we still have one minute. If somebody wants to ask something. I'm going to do myself a question. In your models. Have you mixed somehow course grain and atomistic models. That's a really good question. I'd love to do that. No, we, we haven't. We, we use either exclusively atomistic or exclusively coarse grained. However, my colleague Jonathan Essex at Southampton. He has, he has some hybrid models. I'm not sure of the latest status of those models, but he's been doing some nice work in that area. I guess I'm just, I'm just waiting for him to sort it out so then I can use them. It's like, nice if somebody else solved the problem. Okay. So I think it's time we can stop for a coffee break. Sorry to everybody for the problems with the sound. I will try to solve it before this next session. And see you in a while, seeing you at three for Lynn talk. Okay. Bye. That wasn't really good. Hello.