 še hom občine z med enfinernosti molekularne norespečne simulati. Petr je, da smo videli ideju tem molekularne norespečne simulati, vzifelja na kuheljega, še, da smo vsak bunkot vso občini vse, in so se museli vzwišti numerikali. Tudi sem je veliko guriku, in izgledajte to v nekaj timestep, a danes ideje in prinsipelje za molekularne modeli. Načaj početimo, če je tukaj nekaj aspekt, če je molekularna dinamica? Simulacija. Zato zelo smo videli tukaj timestep, tukaj kondition, in taj držav od tempratu in priču v prvom početku. Zelo, da se pričeš, da je simulacija konditijna. In v drugim početku zelo, da se zelo, kako zelo, zelo, in na konfiguracijenem početku. Zelo, taj početek. Taj početek predaj, kako je taj početek sem simulati. tako, kako je komputacija, je to zelo stejnje, zato zelo stejnje in zelo stejnje in zelo stejnje. Tako, je to zelo stejnje. Na različenem vse možemo simulati zelo stejnje. Tako, vzelo stejnje je zelo stejnje. Zelo stejnje je zelo stejnje. Zelo stejnje je zelo stejnje. Tako, vse zelo stejnje je schovil, in zelo stejnje. Tako, jen 딱 se tem počok je skupnago, in tizje dober je, če je jazel se izgleda, zelo stejnje je zelo stejnje. je složan smelice. But when it's too small, nothing happens and it takes too long before a cure and that's not what usually we want, since we have always the problem that we don't simulate long enough. If it's too large, it causes instability and then we cannot cure immediately, inитivno, kot vsopnike je z vzključenjo zvedanje. Pa v verti, when you choose a molecular model, you have to pay attention to, how this molecular model is set and how the time step is set for this molecular model. Not all the force field for example is the same, it was the same time step. When we have an appropriate time step we can describe izgleden kompromis in simulacija tudi v izgledenih v nostrih sistem. In otroj aspekt, zelo način in atomistih, način in atomistih skupnji, kjer smo vzostajili bombe, uspežimo dve femtosekom tudi. in je div juice dveč, da tudi je zemlječna mojda, s kajim sedaj trenenčem potenšeljem sa vstupném dramatizacijem. Tležito prej dveče je kot srečo relacionira s potenšeljem in z dvečem potenšeljem, da je začalač nerečna potenšelja. Vse nekaj umel Naj nekičnijo starčče nerečno potenšče, in kar da je šaljel potenšel, da je šaljel način. Tudi je to zelo, zato, da kako najmaj glasbeni model da je šaljel potenšel, danes automistik model, so vsega izgledaj šaljel način. In da se izgleda, da se izgleda, da je šaljel način between the molecular model that you use, and the time step that you have to use with that molecular model. There are different ways that can be used to increase the time step. And one I was mentioned already is to apply constraint to your bond. There are different constrain algorithm. Example are links, parallel links, shake. And usually the time step is correlated to the fast motion, and the fast motion is correlated to low mass. So for example, it's very frequent to put constrain on the bond involving hydrogen atoms. Another way is also to use virtual side change. So also here, if we look on the fastest motion, this will involve the hydrogen, but not only the vibration of the bond, but also the angle and the torsion. In particular, for example, for the material group of the amine group. So the use of this virtual interaction side allowed us to remove this fast motion and in this way to increase the time step. How it works? So for example, here you see in gold the position of the virtual side. And so the force is calculated on the atoms of your system, then is redistributed to the virtual side, then the Newton equation is integrated, and then from the virtual side we reconstruct the position of the original atoms. There are also other approach that can be used to increase the time step. One is also to play on the mass of the atom, to make light atom more heavy. But in this way we affect the dynamics also of the system. Then we go more in detail in the simulation condition. So usually we want to simulate, we simulate one molecule, but we want, and this molecule is mimic a molecule that is together with more molecule in, for example, in a covert. So if we look at the baker of an experiment, there are very few molecules that are in contact with the border of the baker. And that is the same that we would like in our simulation, even if we simulate one molecule. And for this way, we implement a boundary condition, and most the simulation are run in boundary condition. What, how this work? So we have a simulation box that is in red, that is our simulation box. Then this simulation box is replicated in all the direction x, y and z. That means that is surround by around 26 other box that are just image of the original box. That means that when one particle is going out on this side of the box, it's just coming in on the other side. This is, so, and this is how we do to avoid to have boundary in our system. We have to define our box. We usually put our molecule inside the box. And the standard idea is that this box is cubic over a tangular. But as we saw before, we have a problem in calculating all the non-bonded interaction, particularly the solvent. So the calculation of the non-bonded interaction, if you think your box is full of water, is computational avie and make our simulations lower. So one option is to try to build the box in a way that is optimized the volume around the macromolec. Usually the macromolec has a global shape, so probably a spherical description is more suitable. So we have been developed a different shape of molecule, of box that are try, that are optimize the volume. For example, for main brain simulation, exagonal box can be a good option since we just have the 85% of the volume of a cubic corresponding cubic box. For a standard global protein, maybe one can think about truncated odecaedric box and that is approximately 77% of the cube or better, orthorobic odecaedric box. That is almost spherical, looks like a sphere. That is resemble the 71% of a cubic box. In this case, since in the edge, as you can see in the of the box, you mainly have only solvent, make everything around will make us save interaction that will save time. So we have to calculate less interaction, non bonded interaction, without losing description of our system. That is also what it is. Another condition that we can define is the thermodynamic ensemble in which we want to run. There are different thermodynamic ensemble. There is MV ensemble, so constant number of particle, constant volume and constant energy, total energy. That is an ensemble that you use mainly in the physical simulation, but not so much in bio-physical and biochemistry. Bio-physical and biochemistry experimented on and then as a consequence simulation, are mainly in perform and canonical ensemble like constant number of particle, constant volume, constant temperature or at the constant number of particle, constant pressure, constant temperature, what is called MPT ensemble. So these are the two conditions where mainly experiment are run. And as a consequence, since simulation, usually are running parallel with the experiment or they have to compensate the lack of information of the experiment, we also are interested to simulate in this condition. That require that we need some approach how to couple the temperature and the pressure of our system. And so now we will see how we can calculate temperature and pressure in MD, what is the way in which we can couple the temperature and the pressure. So the temperature in the molecular dynamic simulation is calculated from the kinetic energy. So the kinetic energy is given by the mass of each particle, the velocity square of each particle over two. And then for the co-partition theorem, we know that this is equal to the number of degree of freedom of the system. And the number of degree of freedom of the system is equal to the number of particle of the system, multiply by three minus the number of cost strain of the system. Then we have kB, it is the constant, Boltzmann constant, and then the temperature divided by two. So if we know the kinetic energy, we know the degree of freedom, we know the Boltzmann constant, we can extract the temperature of that system. So the temperature of my system can be some way controlled by modifying the velocity of the particle. If I modify the velocity, I have a different temperature. So usually the algorithm that we use to control the temperature, we call it thermostat. And the role of the thermostat is to provide us to ensure that the correct average temperature of the system and that the coefficient of the temperature are of the correct size. There are several thermostats, so one of the original thermostat, that was the oldest one, developed in 1991, is the balance with coupling. Thermostat is very efficient in relaxing a system to the target temperature, so the relaxation of the temperature is going in an exponential way. But it does not reproduce the correct fluctuation of the kinetic energy and as a consequence of the temperature. So then it doesn't resemble a good canonical ensemble. In 2007, Bussim developed another algorithm, Velocity rescale temperature coupling, that is an algorithm that is based on balance in thermostat, but has an additional stochastic term. And these ensure a correct distribution of the kinetic energy. So now we have a thermostat that provides not only the correct average temperature, but also a correct fluctuation of the temperature. Then we have another thermostat that provides also a correct canonical ensemble and that is Nose-Uver temperature coupling. This is the philosophy behind this algorithm is different, thermostat is different. Here a thermal reservoir and a friction term is added to the system Hamiltonian. That is peculiarity of this thermostat is that it is relaxed in an oscillatory way, the temperature. So if we are very far away from the target temperature, that might be a problem. So it's not used in an equilibration phase. Then we go to see how we calculate the pressure in molecular dynamics simulation. The pressure is related to the volume of the system and to the interaction between the particle. So if we can see, it's calculated from a first part where we have a degree of freedom, Boltzmann constant temperature divide over the volume and this part is resemble us the kinetic energy. So actually is twice the kinetic energy and then we divide for the volume v is the volume from our box. Then we have a second term and this second term is the virial that describe the contribution due to the force between the particle. And this can be easy calculated in MD when we calculated the force in the phase where we calculated. From this we can extract that the pressure can be controlled by varying the volume of the system and scaling the position of the atoms, of the molecules. So when we couple the pressure, we speak about barostat. We have different way to couple the pressure. We can couple it in an isotropic way. So it means that we couple in the same way along x, y, and z. Or we can couple in same isotropic way. So we couple in the same way in x and y, mind in a different way in z. Or we can couple in a completely an isotropic way. There is also an option that to couple not the pressure but at the surface tension. So here again I go through some example of barostat and as also was in the case for the barostat before is I don't cover all the barostat, all the thermostat. So here also not the oldest but one of the oldest thermostat barostat is the Berenzin and that scale the coordinate and the box factor every step. And that has also the same problem that is reproducing correct the average pressure but it provide a wrong fluctuation of the value of the pressure. So an alternative, a good alternative that was coming out very recent in 2020 Habernetti and Pussy is stochastic cell rescaling. This algorithm is very similar to the Berenzin algorithm but a stochastic term is added and that allow the good reproduction of the fluctuation. Then another alternative is called Paterinello Rama and these thermostat work somehow like nozehover temporal coupling, so I had the term to the Hamiltonian. And also in this case the relaxation of the pressure is go in an oscillatory way. That my cause is that the system is very far away from from the reference pressure, strange oscillatory behavior in the box. So now we go further with more so we saw the pressure and the temperature. How they calculate and we go further with that information on how the system set in. So we saw the condition, so it's a boundary condition, how the temperature, the pressure. And now we go to see how we set the system. Molecule are always in a condensate phase and they are in an environment that is or the cellular environment or an environment that try to mimic the cellular environment to be so close possible to the cellular environment. So that is also what we would like in the simulation. So when we are simulating a membrane protein we set it in a membrane, we would like that the composition of the membrane, the lipid composition of the membrane resemble as much as possible the lipid composition of the original membrane. We will set this, the protein in solution and usually we want to have ionic strands so close, so possible to the experimental one. If it's a protein for example in seawater where we have a 0.6 ionic strands, 0.6 molar ionic strands, we want to have the same type of ionic strands. So we usually add to our macromolec to a solvent, ion, another type of molecule that are present in the experiment. Another, then we enter is another problem when we set the system that we need the macromolec or the molecule that we are interested in. So we need to have original coordinate. This can come from an experimental structure that can be an x-ray structure and an MR structure, cryo-M structure. Or it can also be a structure that is built using for example, homodulin modeling, docking approach, also so some other geometrical criteria, for example, that are usually used, for example, to build DNA or name molecule. For solvent, we also need a starting structure for the solvent. So usually for the solvent we need to build a solvent box, relax, equilibrate. And then we can use this solvent box to solvate our macromolec. These, there is, there are database where we can find that this is one of the database, the protein data bank, where we can download PDB structure. So structure of file with the position of all the atoms of our macromolec. And that we can use as a starting point, for example, and then is what we will do in the tutorial later. There are, can be several type of issue. So not always when we have a deposit structure in the protein data bank, all the position of all the atoms are solved. Some atom might be mixing or some atoms might be just model. So that means that behind experimental structure there is already a simulation going on, but we don't know. So one has always to check the experimental structure and usually to understand which are the really the raw data that come from the experiment and which are the part that has been model to make better the refinement, for example. We sometimes happens that a very key flexible loop maybe is missing. So we have to think about how we want implemented. We can use software like Modeler to model it. And then we have also to think about to relax. Also another issue with experimental structure may be the position of the hydrogen. Hydrogen are not always available experimentally. And all they might be available, but they are specific of the condition of that experiment. And that might not be exactly the condition that you have in your simulation. So that has always to think about. And for example, a PK shift can occur that this might be frequent in the pocket of a box. So it might be the protonation states of the race in that pocket might be affected by the present or not of the ligand. It could be that different atomics states occur and automerism is very difficult to detect with almost in experimental technique. The only technique that provide us some information that automerism is NMR. Then in X-ray structure in mainly in X-ray structure we might have the position of other things like water for example, but it's very hard to know if those are water or armenizium ion because it's almost very difficult to distinguish or sometimes the criteria used to distinguish them is just based on the fact if one can form or not hydrogen bond. But the question are what relevant are those water? Are just you there to the crystallization procedure or are key water that play a role in the function of the molecule? Also in experimental structure to get the structure we might have other extra molecule co-factor, ligand, sulfatane or they can be performed in a special condition that has to promote to see the experimental structure but not necessary will correspond to your condition. So when you take a structure you have to think all of this and also when you interpret your results you have to have an overview of this. And then we have our structure. We are happy with our structure so it means that we have the position of the atom in a file. And probably when we want to start the simulation our system will also have a temperature. So if we start like this it means that we start at temperature equal zero. This is one option. Or the other option is to assign random in the velocity such a way that the distribution of this velocity reproduce the desire temperature. One way for example is that the velocity are taken from the random Maxwell distribution with a kinetic energy corresponding to the design temperature since we know that there is a relation between the kinetic energy and the temperature. So now we have seen in this second part the time step. We have seen the condition, simulation condition with boundary condition and the temperature pressure. Then we have seen how we set our starting structure, the ion, the problem of the selection of the starting structure. Now in this last part of my presentation I want just to speak about molecular simulation connected specific with Gromax that is the software that we will use in this summer school. As I told you, so when we start from our structure that it can be model or take from a database so we have already a priori choose how we want to describe the degree of freedom. For example, here we already decided to describe it in an atomistic way. Then we have downloaded our structure. Probably is a good procedure to visualize it to check if we miss something, if that is looks fine according to what we know. We probably have had already all the missing atom. Then we have to choose the force field to describe it. Since we are atomistic, we will choose one of the force field. We also have to define solvent in the ion so the environment that will be. Then we have in case we have some strange component that are not already present in the force field. We have probably to generate a mean implemented missing parameter and this should be in line with force field definition. And then we have to define in which condition, which temperature pressure we want to run, how long we want to run. And then we have this our starting simulation, starting structure that we can start to simulate. How this is described in term of Gromax so we can see which are the topology file that we have when we use Gromax. We have a specific, we have file that describe the structure, file that describe the simulation parameter and file that describe the topology of the system. So usually when you get the structure from the database for a PDB database, you can run a tools that provide you if this structure is one of the standard macromolecule, it will provide you a topology file. You have to option to choose which protonation states to have some control on the tatoominic states, to the protonation states, to the terminic and other small things. And this, so in this way you have generated your topology file. Then you have this, you have still your structure, but your structure probably is not in solution and you want to get to your structure as a structure in solution. So you will probably have to put your structure in a box. You have to solvate it, you have to add the ion. So to put the correct environment in your box, all the molecule that are required to describe the environment should be in the box. And this we will enter up in a structural file that can be in a growth format, in a PDB format, but they contain the coordinate of all the atoms of your system. Then you will proceed to have to define a file that is called MDP file, Simulation Parameter File, where we have all the setting of our simulation, how long we will simulate, which condition pressure and temperature, for example, boundary condition, all the setting of our simulation. And now we are almost ready. So we have our structure file, we have our simulation parameter file, we have a file that describe the topology of our system. Then we need the tools that put all this information together and you create a new file that contain all the input information. Then this file will be processed by tools called MD Run, and it will provide a series of output, a trajectory where according to the criteria that we have saved all different time point of our system, a file that contains all energy information and a log file. So this is how we perform a simulation, how we have to set the simulation. Usually when we start from the beginning, we have different step in this simulation. Usually when we start from a molecule from the database, we have a first step where we energy minimize, so we relax the molecule in the potential that we have used to describe the model. So we have a first phase of energy minimization. Then we have also a second phase where we will relax the position of the solvent and lion around the macromolecule. Then we will clearly break the whole system and we will move to a data production. And the data production will provide us an ensemble of conformation, from which we can extract a different type of property that can be thermodynamic property, structural property, kinetic project, dynamic property, mechanical property. So we can extract different type of property. We can see that in each of these steps, before going to the data production, we use a different type of NDP parameter. When you finish your simulation, and then you are performing analysis, one other thing that always I say, and then I think it's very important, is to visualize your trajectory. There are different visualization tools. Here I list some of them, but it's very important that you visualize your trajectory because by looking what happens to the atom, you might discover things different at what you thought you want to analyze. It might also help you to set and calibrate better your analysis. With this, so after visualization of our molecule, I thank you for your attention. And I see you in the Q&A section where I hope to get a lot of questions. Any curiosity, anything that you didn't understand completely, please ask. In the future, after this summer school, if you still have question on particular gromax, please go to the forum and ask or look, maybe someone else ask already what you have, you are interested in, and there is already the answer. Thank you very much.