 Well, OK, I'll try to follow up a little bit on your hand and try to introduce you to the GPCRD project. I want to admit that we are coming a little bit from a different community, also applying it to us, and the community which actually focus on studying GPCRD, including Carpenter Systems, you might have heard about. And at the time, we are really interested in understanding the commonality, and of course, understanding is very important. But at the time we were coming across, there was no data understanding, and it was so complicated to compare with other results. We thought we really have to make a step forward and combine a little bit of forces. And in this respect, we combine forces between Europe, within the GPCRD community. We are about 10 labs. Spain, we have Germany, we have Poland, Switzerland, Denmark, and also Sweden. And we have also some support actually from Stanford University. And we came up with our objective that we wanted to generate the kind of first prototype of a database that was able to store simulations dedicated to GPCRD dynamics. And second, without a cake, we also want to make the first effort to generate the first data set. The data set which can be actually submitted to this database. And at that time, we had kind of 91 unique GPCRD complex that had been crystallized and that was kind of a second goal. And of course, the benefits are clear. On one hand, we really promote data transparency and we really try to internet also to promote reproducibility. Other researchers could actually explore GPCRD data and overall it would actually just accelerate GPCR research. So the GPCRD database main features was we wanted to be able to deposit and also to storage data. Then you have to quite provide current tools, of course. Next big challenge was actually visualization. But we were really lucky at that time because Johanna and the people around Johanna had already started the NGL View and DMV server. So it was just the right time to start with that. And we provide some basic analysis too just for the quick view and to get an impression of the simulation involving R&D progression, statistics about link and receptor interactions or also interaction network dynamics which was actually supported from RASMOS and AJ from Stanford University. And extras, of course, we really want that people deposit their simulations there and that they obtain actually really a signable ID like a DOI and also MD, an ID identifier for certations and papers. And last, also that non-MD experts can actually easily browse and visualize the database. So just for you also to understand a little bit the progress of the evolution, we also came up with this idea let's say at the middle end of 2016 and then we were able to come up with the first prototype to the end of 2017. In parallel, we also started the simulation project where we combined the expertise of the GPCR people to curate the structures to set up simulations also to simulate them and currently we are actually analyzing the data and we expect actually to help out our first publications on that. And of course in future we want to expand the server and also kind of initiate a second simulation project on new crystallized structures. So what is really interesting because at the beginning we didn't know where to start so also we had to look around so what kind of information are we going to gather and we kind of created this database using the philosophy of PDB and also of the GPCR data which has been existing also for 15-20 years now. And what does it mean? We really kind of collected very detailed information about the components in the simulation system about the protein, about if there are any mutants and also about small molecules like what is in the simulation like ligands or lipids. The second thing is for the reproducibility that we really try to collect the simulation about the system preparation and the dynamics protocol and finally what is important that you can not that you don't have to install locally any software that you can really browse it without needing any special software. Here I just want to show you a first prototype so which you can actually reach from that website down there. We have first the queries, browse simulation browser for element and simulation browser so you can search for an element which you want to find in the simulation and add it to your browser you can also think, oh I want a simulation from C you add it again to your browser and you can check if there are any hits that match that search. Here we have two hits then you can link on that you can get information to the simulation software source, your memory, timestamp, delta you can also track some vendor data and you can actually download the simulation data. The next thing is to visualize which was really helpful implementing in the earlier as you can see on the left side you can right away play the simulation you can also select what you are interested in so it picks up right away what kind of compotes on the simulation you can show them over here or you can also select in a very specific way by bottom selections or ceiling selections or what we do very specific for GPCRs we know they are very highly conserved residues and motors for different classes which you can do very easily if you just click on it it is very interactive. On the other side we have the analysis which contains flareplots, interaction maps, distance, arms to interaction and which is also very interactive so if you play it you see right away here the flareplot actualizing you can show summarize or you can also then interactively pick on something which will be visualized right away in here a certain connection between a certain transmit where it is for instance. This basically was the first prototype which I think was coming ordering quite nicely just to give in I mean everybody is probably aware of what kind of advantages you have if you can collect data like that within our simulation project we could now simulate all the different types of GPCRs and just to check for instance of if the simulation protocol which we suggested within the community works actually well in the complexity for the ligand stability and if you check you can see for a lot of them it works quite nicely but you can also write it when you come up with cases where this simulation protocol doesn't work that nice you can just go back and see what is missing there so also I could just jump over this so in another case what we actually analyzed I think what is more important for you guys is looking to the submission system the kind of information we are actually collecting and as I told you we are kind of inspired by the TBV so we have actually different steps for the submission which goes information for protein for the small molecules for the complex which we simulate and also for the dynamics information just quickly for the general information for instance the first step protein we get really the unifor automatically retrieved sequences which is important because in simulations you don't usually use the byte a lot of times in the crystal structures there are mutants so you can just click that and actually the mutant sequence here so that we can actually store all this kind of information also for any other type of small molecules because we believe we really wanted to provide a platform which is searchable and with other databases we collect here the SDF files we link them with the demo we have then really kind of different descriptions like the NGP or also other slides which are really easy to search on our browser to come and to find the right simulations which are linked to them then we can actually define if these small molecules are co-crystallized or if they belong to the bulk of the simulation then we of course need and we can upload our simulation files this is important the starting files for simulation that coordinates topology trajectory files simulation parameters other files which include the simulation protocol and in the end we have kind of a summary of simulation components and which you kind of see here an overview and kind of also summing up the number of all the simulated items and also giving specific information about software version and some other things finally I just want to make you aware about the main challenges which we had to face at that moment and which kind of was driving the development of this server the first thing is data reproducibility which we talked a lot before so therefore we really decided to store the starting files of the simulations the coordinates the parameter files and very important simulation protocol so this can be done for any type of simulation if they come from gromats or if they from charm or whatever there's no problem actually let's say format diversity in that sense data quality you can say okay what is going to be uploaded by the people be limited at the moment to if people want to upload to peer reviewed data or the data that are going to be published because we believe there should be a certain quality be provided and so how do you decide if something is going to be published actually what we do is so people are going to upload they can use the link it's not published then they have the private link they can give this link to the reviewer and then once this publication is accepted it will be published the data so if something is not accepted after some point you just delete it this is at the moment just for different ways convenience with the data quality and also space of storage it's like that for we decided on that and what do you do now to make sure that people upload correct files data check up on upload this is kind of really also a little bit problematic which has to be still a little bit elaborated in our case at the moment we just check actually small molecules the coordinates and then the chronology go along as well as for put it means that the coordinates are okay and also the sequence like if you have any mutants which has been indicated when on the submission of PDB files and finally visualization I mean one has shown that already at that point we were facing actually you know how to need special software to display simulations in the browser they have to be able to be streamed the simulation data taking into account the bandwidth and of course also 3D visualization was actually the NGLF we are powered by the VATGL and the ND server so we're really happy about and works very nicely file formats at the moment I mean we can accept different you know coming out from different simulation engines because I mean NGLF we are able to execute quite a lot of different file formats and basically we're depending on that where we have actually problems is what kind of simulation techniques because at the moment we accept classical anti-STMDs and we know this is there are a lot of other possibilities we have we do not know at the moment how to treat metodynamics sampling, replica exchange, etc so what to do with that we are wondering already if there should be just a representative of something because in the end individual frames it doesn't make that much sense to do better the analysis of it so here we still do have problems to decide what to do and with respect to this space and then if we have at the moment as we are quite restrictive so we don't need to worry too much and therefore we can say per simulation we should have kind of not more than 2 gigabytes but that's fine so there's no problem at all yeah with that I just wanted to kind of show you the challenges which we had and how we addressed them and we have a lot of them that were already discussed today and I think let's see if you still have some other ideas and maybe also now with the brainstorm of course I just don't want to acknowledge also the people, the GPCR and the community which have a lot of developing certain concepts and the people that put their hands on to create GPCR and D-database yeah so I hope now the coming sessions will be going to discuss different things questions so just open to answer them so you can also download the trajectory yes yes yes yes especially for the reproducibility the stamping parts and the simulation do you have some strength to do with the cruise for the final job do you have direct links to the trajectories but you can also make like scripting which is yes yes yes you can do that at the moment so we still have the prototype so we have maybe at the moment 30 simulations uploaded because we still have from the simulation project all the other ones a little bit behind because they are analysed at the moment so they will be now actually pushed to be published so are you planning to go beyond GPCRs yeah it's a good question I mean this has been initiated from the GPCR community and I don't know at least for membrane proteins could be interesting maybe because we could talk with some other people to at least share our experience but at the moment it was very specific for the GPCRs because also the visualizer is very specific in our kind of which kind of details you're showing we have a little bit similar thing that now contains all the membrane proteins but I don't know yeah I think we can talk about that we have time for one more good question I think that Tom is actually in person this is super awesome by the way looks beautiful too which is really nice but we should discuss as a community the value of sharing just getting the trajectories in the hands of other people versus being able to enable them to do interactive analysis in the browser because inherently just looking at something wiggle is pretty limited in terms of what inclusion we can make but then you're enabled to do things or you're enabling people who couldn't do things to do that very easily so which one is more important and are there considerations we can have I completely agree therefore we also just have some basic analysis tools for because I think if you really want to analyze you have to do a lot and this is what we provide but for instance also for let's say non-MD experts at least they can look at legal interaction statistics as legal interaction which might be for MD expert on the program or just maybe using their own scripts and doing it in a more specific way but yeah I think the download and the sharing is probably much more important than providing at the moment the analysis there is it more of a tool for finding the data that you really want to download and analyze later than the interactive experience again more than to help people find the right data to download and analyze I mean the search should help you to find the right data you can very specifically say what you're looking for and yeah that's basically one of the goals of that MD server yeah