 Okay, perfect. Can you hear me? Yes. Okay, I cannot see my video, but I hope I appear normally. Yes, yes, you are there. Okay, good. Let me know if there are any strange things happening during the talk. Okay, but thank you very much for giving me this opportunity to talk about SAS 2. And apart from this data analysis part, which is central for SAS 2, I also want to talk a bit about what SAS 2 can do for you and what you can do for SAS 2. As you will learn, the SAS 2 is a community-driven project, so both actually input and output, I should say, is equally important. And so, sorry about that. Okay, so speaking about SAS 2, it's of course important to also mention the techniques that it's primarily been analyzing data from. So I actually plan to cover quite a few topics today, and I learned that's been particularly challenging for these presentations to find the right background. And so I arbitrarily choose some material that hopefully will cover the background of the audience as well. And I did it based on the, watching on the previous webinar. So of course, audience may change this time, but I hope that will provide you enough background to understand what SAS 2 is doing. But by no means I will be managed to cover everything. So I'm going to talk about the SAS, which is small-angle x-ray scattering, SAS, small-angle neutron scattering, and I will also mention a few concepts from data analysis. So let's move on. So SAS, small-angle scattering is the technique that is using either x-ray or neutron beam to investigate the sample. And the part of the beam is transmitted to the sample without any scattering and the part of it is scattered. And when we have the detector long enough, we can then observe the scattering pattern on the detector image. And that's usually measured in terms of the intensity in the function of the Q, which is the scattering vector, which is defined as the difference between the wave vector of the incident beam and the scatter beam. And using small-angle scattering, we can usually investigate the structure at the length scales from 1 to 100 nanometers, roughly. And because of this, we can learn particular information about the system. So we can learn something about the size of the particle, about the shape of the particle, as well as interparticle interactions. We cannot really get direct information about the atomic composition. That would be, to some extent, covered by the next speaker, Andrew Sazanov, talking next week when we will talk about the diffraction. Nevertheless, we can still gain quite a lot of information about the sample. And in some particular cases, when we have the sample, which we call that they are oriented or rented with respect to the incident beam, we can also learn and get information about the particular orientation. And we also can study magnetic properties of the sample. That's what can be done using small-angle neutron scattering. And the spectrum of the samples that can be studied using either sacks or sands is really, really fast. So essentially, it's only imagination that limits what can be studied using sacks or sands. So here is the list ordered alphabetically. Hopefully, you can find something for yourself. I hope your system is covered here. Most likely, it can be studied by sacks or sands. It's just only that I didn't put this on the list. Or someone else that I adopted the slide from, sure. So how sacks and sands are related. So the basic physics or the fundamental physics of sacks and sands don't really differ. But the properties differ really a great deal. So when we think of X-rays, they are essentially really available. And you can use the lab sources to investigate your structure using sacks. As for example, it's shown here, this nanobix device from the Regaku. That's essentially, as you can see, a small device that allows you for the studying nanoparticles using sacks. We can also go to the synchrotrons. And then we have much higher fluxes that allows for the more intense studies as well as the time resolution is improved. Which allows for the sub-milliseconds time result studies. Sample size, that's, there were in the previous webinars, there were questions about this. That's for the X-rays, that's essentially something bigger than 2.5 microlitres, depending also on the setup and the instrument. And that's considerably smaller than for neutrons. And speaking of neutrons, then we have to go to the large-scale facilities to perform neutron experiments. So here is the picture of Loki, which is the science instrument that it's currently under construction at the ESS. That will be for the science measurements. And essentially, as you can see there, that's something probably that would fit 100 of these nanopix devices inside this cabinet. And we of course have to go to the sources like explanation sources as ESS would be or reactor as, for example, ILL is at the moment. The one advantage of the using neutrons is that they don't cause radiation damage to the samples. Also provide to study time, perform studies in the time resolution. However, that's not exactly the same scale as for the X-rays. The sample required though is much larger. So we are talking here about the 100 of microlitres rather than 10s. On the other hand, it also provides something which is called contrast variation, which is the property of the neutrons that allows for the studying part of the complex. For example, if we have a protein nucleic acid complex, we can match out the nucleic acid and observe only protein or vice versa. Match out protein contribution and observe only nucleic acid. And neutrons also have spin. So as have nuclei, so when the neutrons interact with the nuclei, they give the magnetic moment. And by this, it's able to, and it's possible to study, study magnetic properties in solids, both static and dynamic. So neutron experiments usually have a higher entry barrier. So to say you have to sort of make a more justification for the, for the studying them, but it doesn't mean that it's not worth performing these experiments. And as I said, there are some experiments like contrast variation and magnetism that essentially can only be done using neutrons. But in general, these techniques are complementary. And speaking of the entry barrier, we hope that ESS will lower this and will allow more studies using neutrons. So what kind of information do we get in the majority of cases when we don't have magnetism or oriented systems, we usually get this 1D curve. So the detector image that I showed on the first slide can be usually averaged to this 1D, 1D scattering pattern that corresponds to the intensity defined in the function of the back of this scattering vector of Q. However, if we go for this oriented or magnetic systems, then we have something that refers to the 2D images and that's kind of data that we get. So, if we now think of what actually it's being covered by this data. So, this is the protein system that I have been working with at some point, it's called Modulene, which is the protein consisting of the two domains connected by the flexible linker. And we established that there are some confirmations that can be captured that the small angles scattering data corresponds to. But now if you think of the how much degrees of freedom or how or how many parameters you need to describe the systems both atoms and this movement is actually quite a lot. And here we have just the one single curve that corresponds to the system. So it is inevitable that we have a considerable loss of structure information by going from the sample to the curve that we usually have from the small angles scattering. And therefore it's extremely important to actually define your model correctly, also using knowledge from the other technique or physical principles or whatever else we know about the sample. As well as the risk to overfitting to data is quite high. Excuse me. So, for those of you who watched the webinar last week, maybe you recall that Thomas presented something like typical data analysis workflow. And so that's essentially something that we would routinely do for the for the analyzing small angles scattering data. And it works or it's the fundamentals for this are that we acquire the data and then we assume that we get this I of q curve for to the image. And then what we need to do we need to define model with the three parameters. And, and then calculate the scattering pattern for it and match it with the data. So, I will briefly go through the steps now. So the what I mean by the defining the model. So the, this is the probably the simplest equation that one can come up with in terms of the describing the I of q relationship with the structure properties. Of the models. The, the, or copying the models with the intensity so it essentially represents intensity in terms of the density of the particles concentration, something with is called form factor that corresponds to the particle shape and structure factor which corresponds to the particle interactions. And this form factors can be defined differently it can for example be calculated from the directly from the protein structure coordinates, as it's shown here in this example of timer. Or it can be calculated for the analytical model like for example the cylinder of the radius of 40 angstroms and the length of 200 angstroms. On the other hand, the, and then as tractor factor that that is defined differently and we have a different model so to say so for example here we look at something with is called hard sphere. And that's the model that corresponds to the exclusive volume repulsive interactions of the molecules. And the one thing to note here is that in this two cases above we have a maximum and this region we call low Q region, and here is actually opposite so couple this together. And if we have the, if we have them, for example repulsive interaction then we see the effect of this on our I of q curve. So, so this is particularly important when you study the concentrated samples, but in many cases, for example, in the, in the bio sector sense, people tend to dilute systems enough so as of q can be estimated to actually be close to one. So once we define this model, we can perform the bidding. So we let's assume that we have a data shown here and as a blue dots, and then we define the model the cylindrical model that is in this case of the land as it's shown here. And then for this initial step, we can calculate the property which is called chi square, which reflects the discrepancy between the points that we get from the experimental data and the and points calculated from the model. And this is divided by the error. And so this property is usually used for the optimization we are in this sense that we are minimizing this value in order to get fit. And once this is done, we can get the refined parameters and as shown here this data being corresponding to the cylinder model of the 440 radius and as you can see the drop in the chi square is going to struggle. So one thing to note here and if you recall what I said a few slides back I mean the risk of overfitting is high so one has to be careful in terms of just using the chi square optimization, especially in the system that involve a lot of degrees of freedom. So let's go back to the slides that I showed before and we'll now segue to SAS view because SAS view it actually covers all these three points from the model definition to fitting and producing the results and nice plots and whatsoever. Very briefly about the SAS view, SAS view originates from the National Science Foundation founded project called dance that was initiated in 2006. And it continued like this for the few years and then in 2013 it was turned into a community driven project. It's currently supported by nine facilities both x-ray and neutrons. We have about 40 contributors, maybe even more. About 15 are active at any one time. And we have a steering body consisting of pole battler from NIST, Matthew Ducey, O'Rannell, and the Jackson ESS and Steve King Isis. And the most coordination of the project is actually done through the bi-weekly calls when we talk together and decide what we do and what we not do and have fun as well during these calls. So that's really lively collaboration and nicely going forward. And we also have regular camps and other events that I will mention later on. And most importantly, some of the SAS view members are on this call, so I am at least aware of the pole battler Steve King and Nicolas Martinez that said that they will join this call and they can also participate in the discussion if I need basically a backup on some questions. Because that's really one of the strengths of the SAS view, that we're coming from the different fields of the expertise and we work together in this interdisciplinary environment and I hope we are producing a good piece of software. So this is how does it look when it comes to program and that's how actually our day-to-day work looks like. So I think that there is a typical understanding of the SAS view as being a primary coding project, but I would say coding is just a part of it. I mean the results of the part that involves working on the theory all start regarding the maintaining the infrastructure project management documentation and tutorials are actually equally important as all the other parts and educational outreach is also important as that we try to perform. So coding is not only part of the SAS view project. So now let's dive into the SAS view. So this how SAS view looks like when you open it, at least on the Mac OS X, and then you can see that you have some load button and send to the fitting. And I will go through some of these functionalities. Now, so this is how the typical fitting looks like. So as you see, it's become a little bit more easy. So I will go, I will try to go step by step for these different functionalities. So first of all, we can load the data and we have a data management table that allows for the loading common data formats, including an ex-cancer format as well. And then we can what we call send our data to different analysis so one can choose between fitting P of R inversion correlation functions and invariant calculation. One can also manipulate plotting to some extent in this in this field. And then speaking of these models, I mean the model definition. So what I what I presented before was, for example, the cylinder model, which is one of the 70 models of the form factors, or even more than 70 models of the form factors that we support in the in the SAS view, so they can be choose from this different menu from the categories and modeling as and the same is true for the structure factor, which can be coupled with the form factor. Then we can define something we just called polydispersity. So in the real life, we usually don't have samples or the particles in the sample that look exactly the same, but they may differ slightly in the overall shape. And therefore, we may want to introduce the polydispersity of the data and SAS also allows for this. The same is with the resolution smearing that's more related to the to the fact that how the experiments was performed and the instrumental setup. And SAS provides the two ways of actually defining either manually or automatically from the from the file if this is of course written to the file. And when it comes to the feeding modes, we can actually perform either single feeding with the just the one curve as it's shown here. But we can also do the do the batch feeding, which means that we can apply the same and feeding of the same model to the multiple datasets, as well as we can perform simultaneous feeding, which is the feeding that involves using the common parameters across different models for the for the for the multiple datasets. And I also mentioned that we can perform this to the analysis and in a SAS with the formulation from for this is done with the something with this called orientation of polydispersity. And it's, it's been set up in a way that it decouples the frame of the orientation from the from the form factors of to say, and the, and therefore it can be also performed efficiently. And as a matter of fact, this, this computations are quite computationally intense. And therefore, we particle are for this models, but not only for this but also for others enable computation on the graphical processor units, this graphic cards, and then allow for the fast execution of the feeding. The other feature of SAS view is that it can use something which is called CSAN's data with the spin equals small angle neutrons category. And that's very interesting and technique that's using polarized neutrons to actually extend the sort of the this limit of the of the landscape that can be pro by by some so here we are talking about the range from the 20 nanometers to 20 micrometers. And SAS views set it up in the way in such it set up in the way that it can actually use the full potential of the form factors that that that we have readily available and there is a trick mathematical trick called uncle transform then allows transformation of this models into the CSAN's data. And it's, I'm, I'm not going very much into the details about this technique, but if you are interested, then I would, I would refer to the papers of the Dean Bowman, who been also contributing to to SAS view on this project. And the, as I mentioned them. And the sort of core of the optimization is, is driving the way that the feeding is performed in SAS so it's very important functionality and actually for SAS we benefit a lot from the project called bumps, which is developed by Paul Kinzo from this. And it allows both conventional and based on optimization, which provides uncertainty estimation of the input parameters and that's very important in in terms of the small angle scattering because you actually, as I, as I pointed out that the loss of information it's it's considerable. And therefore we usually have the quite considerable error bars and therefore we also want to know how this uncertainty uncertainty is reflected on our fit. So bumps allows this kind of analysis as they are shown here at the different plots and the trajectories as well as uncertainties, as well as provides quite some choice of the of the optimizers. One other nice feature of SAS view that I think it's a particle and useful for the, for the community is the fact that it allows for the easy addition of the packing models. So if you have the system that you cannot really choose this any analytical model from the 70 plus models that we have in the SAS view, then you can write your own. And that's actually this task can actually be done in both ways. So one can start writing up directly from the Python and then looking at that, usually looking at how other models are structured, but we also provide the tool that the model editor. And that allows in the few simple steps to generate the Python code that can be then used in such and that's actually directly available in such and you can use it straight off. Once you click apply button on this model editor. And if you then have some complicated model and then actually requires more speed optimization than usually in the recommended way is to go for the for the C implementation that is also nicely interface to Python that we support so it's actually can and so says we can support two types of model and model files and and performance and optimization on the GPU. It then also ready and the and the feature of the models that at least from the from the SAS view and models that we support we provide is that we, and that we always have a full description of what's happening in the in the model so it's easy to keep track on what exactly the math is that we like to perform. And we also test our models so we check in the every time we add a new feature to the code that actually the model keep the same answer as we would expect. Okay, so I've been now discussing mostly things that and then assume that you know the model but that's not essentially not always the case. Sometimes we, especially at the initial stage of the analysis want to do something directly from the data. And this can be done through, for example, calculation of this poorer that scattering invariant, which is essentially integral that it's proportional to the fluctuation of the scattering of the face composition. And what it actually provides it provides the independent estimate of the volume fraction that then can be used to to understand porosity of the material. And surface area, which can, for example, corresponds to reactivity. It can also be used as a sanity check if you want to sort of see if the same sample gives the same answer of the different instruments. The other possibility is to use the P of R and the per distance distribution inversion. And this is probability of finding a vector of land are between scattering centers. And in SAS we use this P of R formula with this using Morse derivation. And what P of R provides it can provide the information of the, on the maximum dimension of the of the molecule. But as you can see here, I mean, depending on the shape of the particle. We also get the concern of different P of R functions. And therefore we can learn something about the our system, even without introducing the model so this can be done directly from the data you can calculate this property using Fourier transform, or inverse Fourier from scattering data. And in SAS this can be done also through this what we call this perspective. And the, and this, there is a part from the, from the, from the fit P of R fit that we can generate. We can also use the exploratory tool which shows us where this D max should be sort of defined because there are some parameters that one can have to set up in this analysis and SAS is providing some nice features to explore this values. The last sort of core functionality, as I would say, in SAS is something that is particularly useful for the material science and that's the correlation functions that can be interpreted as a imaginary rod moving for the structure of the material. And SAS provides the different correlation functions one and three dimensional correlation functions that can give us information about the periodicity or, for example, if material is amorphous. So, so far I've been discussing this analysis or the perspective functionality of SAS view. However, that's not all actually that the number of tools of the future is actually much larger. And just to mention the few we have tools for the scattering and density calculator, slit size calculator, generic scattering calculator, for example, to calculate the, and the scattering pattern from the, from the protein coordinates. And, and also a few other utility tools. So I'm not for the interest of time I'm not going to get into these details but if you're interested please check either in SAS view or other resources. So very briefly about the SAS architecture, how is it structure and what one can do. So the part that I was talking about that was the sort of outer layer of SAS view, which is this graphical user interface and that's what we as a, as a user scan interface with that with the program without touching any code. However, these days it's becoming more and more increasingly popular to use the Jupyter notebooks and other scripts to to run the software. And actually, SAS view also provides them means to do to use this kind of interface. So we have something which is called SAS calc, which is essentially the backend calculator for the graphical user interface. We have this package which is called SAS models, which is defining both form and structure factors. And we have this thing called bumps, which is the optimizer. And because of this, we can, as I said, run SAS view from the script. So here is the example of the feeding using SAS models and pumps. So I'm not really touching the outer layer at all. And once this is done, one can detach the or the deploy the script on the on the computer class and rabbit run it over there. The other possibility that we also been working on, I mean, though it's still a little bit experimental is to define all your parameters inside the SAS view and your feet, save the state to touch from the GUI and then run this sort of project file on the on the cluster. So that's also the need functionality that doesn't require any knowledge of the scripting. However, it opens up for this possibilities of using computer cluster if you have a long analysis. This is just to illustrate the another example using the scripting interface. So we can also use this for the for the bird distance distributions calculations as well as we can just use the SAS models directly to generate our favorite form factor or whatever. So this is this is essentially covering most of the of the SAS view. In terms of the functionality, what I would like to briefly talk about is how to use SAS and what resources are provided. So this is typically the how you start using it you go to the to the SAS view website and then you choose the choose the downloadable version, and then you start interacting with this as you probably notice, I mean, we currently support version four and version five. And version five recently is being getting more and more attention so there are more frequent release. And that's something that will eventually move on, but there are still user users that prefer use version four so we still support it. If you don't like good at all, no matter if four or five, then you're of course very welcome to clone the repository and get an interactive code code is written mostly Python, plus some C part that it's used for the for the GPU optimization. But you're also welcome to do so. And so I mentioned that the five version is being released more often, often these days, and we mostly been releasing the point releases that will be by point me release I mean the 502. There is actually 503 coming up soon as well, that we've been fixing many bugs, but also providing the new features and improvements, mostly driven by the user requests. So if you want to get some feature done, and I think that the best and some feature done sooner than later than, then you're of course welcome to to talk about this and or report it to the different things that I will just discuss on the next slide. And for the sort of bigger picture we have something. We have a SAS view roadmap which is the living document that we keep in the five years span and that defines more top level goals for SAS view. And here are examples of them. Of the what we've been planning for the for this coming year essentially. So as I said, we, we plan to release a bigger version and and complete some, some other milestones. So if you're interested, then please take a look on the on this link below. However, if you, as I said, encounter some issue and you want to share it with the community, you of course are welcome to email us. But also, we store everything on the GitHub so there are no secrets, everyone can see what we currently been working on. So day to day issues are stored on GitHub and are available. If you're interested in this, then please take a look on this. And I would also like to mention what says you cannot do and it has to be for about it. So we are not saving the word here. And so we don't do molecular modeling and as well as ab initio modeling. So there are, I just mentioning to software packages here. So SAS see that's been doing molecular dynamics and combining this with a small angle scattering. So if you are interested in this kind of studies then I would refer to to these approaches. And if, especially if you are coming from the bio sun sucks or suns community, you most likely encounter software called at SAS, which is really comprehensive toolbox that is also providing many other functionalities like this ab initio reconstruction that I would advise to use with caution. But that's, that's not really functionality that we that we cover with SAS so as the other rigid body modeling using the protein models. That's to some extent true with this with this molecular modeling because I also I will now demonstrate on the example that I worked. And that now with this with this possibility of using SAS from script we can actually achieve many of this functionality as well. So this is the example of the virus capsid assembly. So the, the, the project that I've been working on at the Lund University. And the basic idea of this project is that we have a self assembling virus capsid so these are proteins that assemble to the, to the final product this wall called capsid. And typically what we don't know is what happens in between. And if you think in terms of the number of combinations that you can have as for example this capsid concept of 240 subunits, then you have a really humongous number of the combinations that you can go through one state from this from this start state to the end state. And we combine molecular modelings with this with the small angle x-ray scattering and collected the data at the ESRF last year that was done in the collaboration with my colleagues from the different institutions and for the trained eyes you can see that for the, for the later time point for the trained small angle scattering eyes I should say you can see that you have a portion like structure coming up at the later stage. So long story short we established that the disassembly pathway proceeds truly essentially to intermediates so we have a dimer state at the beginning, then we have a hexamer they are not exactly on the same scale. But then we have the, then we have the intermediate that corresponds to the 80 subunits and then at the end we have a mixture of the different capsids and then using the some clever mathematics we established that and that we have a, we can infer the population weights along this time. That's an exciting study but the mostly the reason why I brought it up is that at the beginning of the studies we didn't really know what is the exact composition of the end state and we've been getting information from the electron microscopy that it's, that there are, it's mostly dominated by this bigger capsid. I consider the 240 subunits and what I did, I combined this atomistic model together with the SASV analytical model in order to learn what, what are the fractions, what are the parameters of the potentially the other models and based on this we established that we can actually, or that we have the mixture of the, of the different capsid which is smaller than the, than the one that we are getting an electron microscopy data for. So that helped out with the initial analysis to, to establish this fact that it's consistent in this data. So just very briefly finishing off because I'm running out of time and we provide a lot of resources for the education and outreach. And so we have a, of course, website documentation, different tutorials, we've been teaching different schools and giving different courses. We also have a eLearning platform. So the, differently, we, the SASV course is also ported on the, is available from the enutrons.org which is a eLearning platform. And we also have some other communication channels. So please check them if you're interested in them. The two aspects that I want to mention are the one is Marketplace, which is available at this URL. And that's actually very nice venue for the, for the community contributor model. So if you develop the model that you want to share with the rest of the SASV community, then you can upload it over there. And that would be available and one can use it for the, together with SASV in the different, in the different aspects. And, and that was actually work done by the ISIS summer students to be the archcruel who did an excellent job on setting this up. And that's really a nice tool for the community. And I mentioned this cold camps or solitary camps and other events. So we are planning to provide more training. So we have a sort of long-standing goal for this, but we want to have a, have a boot camp, which would provide the training from the basic use to becoming essentially a core developer. One can stop at any, any, any stage one wants. And we also, and if you're interested, then there is a link to Syllabus here, what we're planning to do. There is no dead state for this, but we will announce for this as also for the other events, all the relevant information and these different channels. And, and we also recently been having hackathons because we have to move our regular events to the, to the virtual events, but I think it's been working quite fine actually. And we, and we are still hoping to have a cold camp 10 at some point soon. That was planned originally for the April this year, but it was canceled for the obvious reason. And then most likely will be held at Caltech at some point. But as I said, I mean, please follow this, this communication channels if you are interested. So lastly, I just want to say that you are very welcome to come and work with us. We have a lot of fun together. We don't only do the coding theory, cracking and math calculation. We also enjoy a simple pleasure of life and like eating and stuff. And with this, I would like to thank you for listening and I'll be happy to take any questions now. Thank you very much. My take was a very interesting talk. So we have several questions for you. Hope you're ready. So we start with a short question from one of the attendance. So does the form factor also includes the size? That's a very short and general question. Let me open the chat. I want. Okay. I will also just try to read. So it's like, so do I understand that the form factor also includes size? That's the question. Right. So, okay. So I was trying to just also follow this. So sorry about that. No, it's not the difficult question. And so yes, in this sense that you, when you define the form factor, then you account for the size and shape of the particle. Okay, so we continue with other questions. So can we calculate the pore size and shape of a pore material or only it can be measured for single crystal using stocks? So both stocks and science can be applied to the porous materials. So that's not, that's definitely possible. Okay. The next question is, Susview, could find application in organic thin film applications? Can you repeat this? I mean, there is something with the noise. Therefore, I've been trying to find the chat. Yeah, you could build our microphone. So Susview could find application in inorganic thin films. Inorganic films. Inorganic thin films applications. Well, I mean, I personally don't have any experience with this systems and I couldn't recall any paper doing this. But maybe my Susview colleagues can comment on this. Paul, Steve, Nikola. We can unmute them if we find them. So Paul is with us today, Paul Butler. Yes. If they want, of course, other than that, I would say that it's most likely yes, but I just don't know. So let's welcome Paul here. He may still... On the audio. On the audio. But you might not be... Okay, we keep... Okay, Paul, you need to unmute yourself. Good. Welcome. So nice to have you with us. Are you in the US? Yes, I am in the US. I did not fly over for this. Okay. You're welcome. The answer I'm trying to think on thin films. If you can model it, then SAS view would be used for it. I try to think of thin film applications. Mostly what I know people do with that would be usually looking for domains, maybe, in thin films. In which case, definitely, you could use models to try to understand perhaps the size of the domains. If you are looking at texture, then you might be using the invariance to look for the interfaces. So yes, if you can do small-angle scattering with it, then you can look at that data with SAS view, essentially. Great. We keep receiving more questions and more detailed questions. So when Paul turned up online, people started to get more curious. The person in the audience is asking, I encountered some misbehaving when activating polydispersity. I will try the latest version first to see if bugs are solved. But can you comment which procedure is expected to communicate with the developers when encountering some issues? Yes, thank you for this feedback. We are aware of the problems with polydispersity, especially for the previous versions of SAS view. If you are using SAS view 5, because I presume that's the one I think for the SAS view 4 there shouldn't be. So if it hasn't been solved with the 502 version, then hopefully it will be solved with the 503, because we have also been addressing polydispersity issues over there. So I would recommend checking out one that will be out soon, hopefully. But the way to interact, the most straightforward way is to drop us a line and write an email. And the email for this, it can be found on the website or in this presentation. And the other possibility, if someone has the GitHub account, one can also create an issue directly. And by going to the interface and then writing, I discovered this and that and so on. And also the GitHub may help out actually recognizing that if this is common issue and actually we've been already working on this. So that also may give you a hint that okay, they already know about this and they are working about it to resolve this. Great. So we want to remind the audience that I can always send you questions. Yes. Can I add one thing for one second? On the polydispersity question, we know actually it's not fixed in 5.0.2 completely, we know that. Nikola, who from ILL has been working on that hackathon. And I believe, having looked at it, that it will fix everything that we now know about. So if you wait for 5.0.3, which should be out in a few weeks, I hope. If you still find problems, then either send a note to developers at sasview.org or on GitHub. That's great. So yeah, the audience acknowledge you for the very detailed answer. We have another question. Does sasview provide solutions for size, spacing, correlation, approximation for dense packed particles with certain size distribution or polydispersed orientation in loosely packed system with unique shaped particles? It's a lot of questions I can repeat. Sounds like very specific, domain specific question. I would say that probably would require... So we can provide what we can provide and we do our best in providing this. We cannot support all the models possible. So to be honest, I don't know exactly if we have a structure factors to model these interactions. But the other part of answer is that nothing prevents from writing it. Hopefully the mathematics formulation is done then and can be done through the plug-in model system. Yes, exactly because we want to remind to the audience that I can always find all the speakers information online on the links web page. So they can always contact you, the developers and then maybe ask and interact with you because you're very kind and so available. So we really, yeah, thank you for this availability. We continue with the questions and those are more general, I would say. So can we use sasview for 3D topography? Not that I know. Maybe in some aspects. Yeah, simply I don't know. I wouldn't say directly but there might be some aspect that will be complementary to the other software. Then there is another general question. Any preliminary characterization tool before we use sacks or sensors. So because you have been talking about your data with certain parameters. So before going actually to the real experiments can be done before home. Yes, of course, especially if you measure at the large capabilities, it's very important that you have a sample that it will behave and you can perform the measurements. And you can come back home with the good quality data. So I personally have experience with the biomolecules. So I mean the anything that can tell you about the concentrations and the oligomeric states that usually useful like AUC. For example, I don't really know how about the other materials, what what can be applied. But I usually the if you have a monodisperse sample, especially for the for the biosystem that that usually helps a lot. And for this I would advise for the following the instrument specific documentation. Usually they provide the hints what what you what you should expect or the or the or what are the sort of the checklist that you should do. Before going there just not to come out disappointed. I mean, we can do our best, but the data has to be good enough. I mean, the information lost is considerable. So we already working with the difficult problem. So any other factors that it's differing the data is of course just making problem more difficult to solve. Yeah, but it's really interesting to see that people can actually be introduced to the software that will be used for the data analysis. And then when you know what are the critical parameters and focus on them and get information about that before going. It's really important. So we continue with another a little bit more specific. So how do we calculate the surface per volume ratio from scattering vector? Sounds like an invariant question. Maybe Paul can take it because he recently been cracking the match to the to the death for this. So I think you are the expert on this now. But you can maybe keep it keep it general and maybe people are really interested. So I think the the best answer that I can give. I mean, if once the this 503 version is out, Paul did an excellent job on putting the commendation for this. And the everything is very nicely written over there. So I would say that for the particular equation that I would recommend getting this. And the yes, because otherwise I'm not sure how I can go over the map simply, but of course. No, the math is complicated. You don't want to look at that. Basically, it's very simple. If you're wanting to look at the surface to volume ratio, what you're really looking at is the interfaces. So you want very good data at high Q. That's where you're going to get all the interfaces will show up. So if you know the porode constant, which you can basically that's where you get the extrapolation high Q. So you want good data at high Q. And then just plug in those numbers into the invariant calculator and you will get the surface to volume ratio. So you want mathematical details, you can read it more carefully. Very nice answer. Thank you very much, Paul. And we have the last question then. So can we estimate that the assembly. The state of micromolecule system in solution. So. The question is using science or the other methods. So I mean the. Yes. And the answer is generally yes. So I mean, that of course it's sometimes tricky because you can have the convoluted signal. So essentially if you have a mixture of the different oligomeric states, then what you see from the scattering does the mixture of this. So we don't see this for particular species. So you have to the convoluted signal and there are different tools to do this. I mean the SVD is the singular value. The composition is one tool. And there are also the potentially better approach using, for example, MCR analysis, not getting into the details. But one can also do some other decomposition and I've been developing for this time result. The system, the capsaid assembly, I've been developing the basin approach to decompose the system. So. Currently from SAS, it's not directly available. Combined together, this can be done as I demonstrated in the example that I showed. So in general from data, yes, from SAS view, not at the moment. Okay, so thank you very much.