 Good afternoon everybody and welcome to the BioXL webinar number 69. So today we have at an hospital from the Institute for Research and Biomedicine Barcelona. And he will tell us about BioBB workflow and BioBB API in an integrated web based platform programmatic interface for biomolecular simulation workflow using the BioXL building blocks. Building blocks library. I'm most in this webinar, I'm Alessandra Villa from PDC that is hosted by the Royal Institute of Technology and Stockholm. And with me there are Otto Anderson and Marina Bocciano from CSS from Finland IT Center for Science. So I want to inform everybody that the webinar is recorded. And during the webinar you have the possibility to ask questions using the Q&A panel. And this is, you find it at the bottom of your Zoom application depends on which operating system you have. You can see this symbol or you can see this symbol. You just click and you type your question. At the end, we will read the question for everybody. And Adam will answer to your question so you can type the question whenever you want. And I usually suggest to type during the webinar not wait at the end because maybe someone forgot and it's always good to have the day. So the reason why we read the question is because not everybody might be able to see the question and the speaker will not answer to the question that is written we just read it to the speaker. Or if you have some question or curiosity after this webinar, you're welcome to join the forum, the BioXell forum that is located hosted by askbioxell.eu. And they are asked questions to Adam. There is a special category that is supporting the bioBB building blocks. There is something about Adam. Adam is currently a postdoc fellow at the molecular modeling and bioinformatics unit that is hosted by IRB in Barcelona. But in the meantime is also a software, a research software engineering of the Spanish National Institute of Bioinformatics. Somehow Adam managed to combine a computer science background with a PhD in molecular modeling. And then it will give, then it will allow you to get close to the bioinformatics world, and then are fascinated by this field, in particular of the structure of bioinformatics. It is involved in several scientific projects, both on RBI, but also on the Barcelona Supercomputing Center. And is currently leading the workflow team in the BioXell Center, a center of excellence. We have developed and coordinated a set of public web server and database. Most focus on molecular structure of flexibility. And ENB integrated platform is one example. And now I'm happy to listen to what he will tell us. So just give me a moment. Please, Adam. Thank you, Sandra, for this nice introduction. Let's see if I can share the screen. Okay, see if everything works a pointer. You can see the pointer. Yes. Perfect. So thanks, Sandra, again. And welcome all the attendees. Thanks for being here in this new webinar series, BioXell webinar series. Let's talk about a couple of projects that we have developed in the ecosystem of the BioXell Building Blocks Library. The BioBB workflows and the BioBB API. The first one is a web server. The second one is a programmatic interface. Both of them interfaces for biomolecular simulation workflows and both of them working with the BioXell Building Blocks Library. I've divided this presentation in four different sections. I will start with an introduction about the two different projects that I'm going to present today. I'm going to tell you a story about when all of these started. I also will give you a very brief overview of the BioXell Building Blocks Library and then the main objectives of the projects. Then I will go through two different projects, the programmatic interface and the workflows web server, and we'll finish with a bit of concussions, summary and knowledge and questions from you. So starting from the beginning. I would like to start with a story, which is basically a piece of my life in research that started in 2008, as you can see here, that's the period that goes from 2008 to 2012. And at that moment, I started working in this National Institute of Bioinformatics in Spain that as you can see here, it was built up from many different nodes around Spain. And basically we want to coordinate integrators, you can see here, bioinformatics resources in Spain. And as we were different nodes throughout the Spain, we wanted an architecture, an infrastructure, bioinformatics infrastructure able to coordinate and to interoperate with all the tools that we were developing at that moment and we decided that we could use web services. Web services at that moment were very popular and were basically little pieces of software, very small pieces of software that were running in the service provider. That means that all the different nodes have developed different web services that were running in the different nodes. We published these services in the discovery agencies that you see in this plot, and the user finds these web services through the discovery agency and then interacts with the service provider to run the web service. It was very nice at that moment because this was fantastic to do this kind of things, what we developed a set of web services in the different nodes, and if we joined these several little pieces, these several web services, we could build workflows like this one, which is a very complex workflow that that moment that you can still find in my experiment. And that is missing microarray data as an input, a statistics using our software, database connections, retrieving information from different databases mixing all of that in one single workflow and as I was telling you, you can still find this workflow in the my experiment website. This is a screenshot of an Apache Taverna, which is a workflow manager tool graphical user interface tool that we were using at that moment. That was fantastic for us. The question that we had at that moment, because I'm, I always have been involved in molecular dynamics simulation in structural bioinformatics field, and in particular dynamics and flexibility of proteins is good. So this kind of web services for these kind of methods, which involves very big structures trajectories, big data that is going up and down the network. This is just an example of what we wanted to do so we at the moment we were interested in automating the process of preparing a protein solvent system running the molecular dynamics simulations and then analyze the resulting trajectory. I can answer you the question and the questions that the answer to the question is yes, we were able to do that. That was not easy. We took us four years to develop all these web services that but that is you can see here we developed more than 100 different web services at that moment. These are the two molecular dynamics simulations and this is just an example of a preparation of a protein from the PDB website to the proteins are random by water molecules counter ions and so on. So the system prepared to start running a molecular dynamics simulations. And basically we use the same schema I was showing you before so this is our group the service provider where we run all of these different web services and the discovery agency was the bioinformatics the user found all the services through the IMB and interact with our tools here. As a curiosity here we developed all of these using an ontology of data ontology at that moment was pretty new that was called movie and bio movie as you can see here and if you see the name here. The name is Mark Wilkinson, which is basically the guy that revolutionized the data world with the fair data principles, 10 years after this. So, the way that we use these web services, where using this taberno work from manager that I was showing you in the previous slide but also in a scripting way at that moment pair was famous more than Python. And this is just an example we took an object from this bio movie ontology, we built this object with a PDB code, we downloaded a structure from the PDB database, and this structure is saved in a new object that it's a PDB object from the bio movie ontology and here we can retrieve the content of this object all of these web services and the scripting were called MD movie. And then, after that, we also developed a graphical user interface on top of all of these services and this, of course, was to help even more to ease the entrance to all of these the usage of all of these web services and basically in the web. As a web server, you could take one PDB file and work with a PDB file launching all of the web services that were available and also preconfigured workflows using these web services. And this is the publication of the MD web and MD movie that was 2012 and as you can see here in this plot, these are the statistic user just statistical server we have maintained the web server. These are statistics from 2015 to nowadays. And as you can see, and especially the last four years we have had like almost 1000 different new users per month. What this is telling us is that there's still a lot of interest in this kind of interfaces this kind of graphical user interfaces to run molecular dynamic simulations. If you want to know more about the MD web, we had a special webinar number three in bio Excel at the beginning of bio Excel about atomic dynamics using MD web. Now, let's jump to the 2016. And I will tell you a bit about the bio Excel building blocks. So in 2016, one very important thing happened. So there is the building of the bio Excel center of excellence and I'm sure that you all know about this because you are now in the webinar, organized by bio Excel but basically by Excel is a central hub for by molecular modeling and simulations if it was a horizon 2020 funded project now is a HPC. And that these are the partners that are currently involved in the project we are in the third edition of the of the project. At the beginning of bio Excel, we decided to think about a way to use all of our biomolecular tools that we knew at that moment and try to make them interoperable. Basically the bio Excel building blocks library these pieces of Lego here that represents functionalities of different tools, we just grab these functionalities in Python code, and then these, these pieces are very easily integrated and generating with these workflows by molecular simulation workflows. This was of course then you can launch and manage this was with different workflow managers, but the main idea of the building blocks is that we make our tools interoperable between each other. So those are the categories we basically gathered together different functionalities of the bio Excel building blocks in different categories so for example everything related to Gromax is in one module one category that is called by the big Gromax same happens with number modeling tools, chemo informatics functionalities, virtual screening, protein, legal interactions, pockets, quantum mechanics and so on so forth. And all of this is available through different packaging systems or conda, for example, Docker containers and singularity containers so by a big Gromax is available in one single conda package also in a Docker container and also in a singularity container that means that all the dependencies that you need to run all the different tools that are inside this by a big Gromax all the different bio baby tools is integrated all the dependencies are already integrated in these packages. For example, we take one of these which is the virtual screening one. These are all the different building blocks that are included inside this by the virtual screening category or module. You can have here in that this is the screenshot of the main website of the bio Excel building blocks you can find the GitHub repo with all the source code. Read the documentation with all the info that you need to run, install and run the building blocks, and then you have the conda package the Docker container and the singularity container so if you download that or you install that in your machine you will have all these singularities together with all the dependencies needed so autodogbina will be installed by default inside these packages, and also in this case f pocket. Same happens with another example which is a Gromax one maybe you are more familiar with this, all the different soup tools that are integrated in the Gromax and the engine package. Okay, apart from that and as we are interested in workflows today, we have a set of demonstration or transversal workflows workflows that we are using in our everyday life such as for example, the preparation as I was showing you with the of a protein for molecular dynamic simulation so taking a PDB from the PDB database and building all the system with water molecules and counter items. So this is a workflow generated with the bioso building blocks and you can find it in the website, and you can find it also packaged in a conda. In this case in a conda environment, it means that we can use Python and a Jupyter notebook we just generate the environment with with conda and then all the dependencies for the workflow will be already there in your computer to start running the workflow so take a look if you are interested to the website there's a lot of information there we, our workflows are also exported to common word for language and galaxy but we don't have time to talk about that today. Last slide about the introduction on bio baby you have also the publication here and just a bit of a curiosity about the number of downloads, again, as in MD web there's also a big big interest in this kind of tools to automate the molecular simulation workflows, for example, the preparation of MD analysis or the analysis of MD trajectories and this is the example of the baby analysis package and the number of downloads more than 20, more than 30,000 downloads actually for this specific bio conda package. Again, if you want more information about the bio building blocks there's also a special webinar series in bio Excel that you can look, you can watch in YouTube about the bio baby library. Okay, so the aim of the project is basically to put together these two different things so that the bio baby library that I have just introduced to you, using this library to update or even replace the old MD movie and MD web projects that I was showing you at the beginning of the production. That means that we could use new program versions. This is very important because the versions of grow max and MD different MD packages that are MD engines that are integrated in this MD web are really old nowadays to extend the functionalities. For example, as you've seen in the modules bio baby modules, we have the possibility to use virtual screening machine learning and different different new tools and user rest API programmatic interface instead of the old shop web services that we were using at that moment for MD movie, and also use the new technology that exists nowadays that in 12 years I can tell you that is completely different from what we had in the year 2010. And in general, we want to offer easy access to buy more across simulation processes the same idea that we had when we built MD movie and MD web. So these are the two different projects and I'm going to show you now and I will start with the first one, which is the programmatic interface to all the different building blocks integrated in the bio baby library. So the concept is very easy. In a way if you want is exactly the same as we did with the MD movie project, as you can see here so these are what services this is the rest architecture which is pretty the same. So you can access to a rest API remotely through a computer, you have different ways to access this data this data is basically most of the web services or the rest API are stored in a database and the rest API is the same information from the database and giving the information as a result to the user in standard formats like Jason or XML. Important is that the processes are still running the providers infrastructure, of course with limitation if you think about more like our dynamic simulation we are not giving MD for free computational resources for free. So the very important point is that there's no need of any installation in your own computer because everything is run in the providers infrastructure you just need the connection to the internet. So the rest API is basically at a collection of endpoints and an endpoint is a digital digital location like for example a URL is the typical one so this for example is a great example. So what we're going to talk about the rest API is the PDB in Europe rest API. Here is an example of a URL if you click on the URL. You will have all my slides available after the after the webinar but I'm sure that you are familiar also with this if you if you click on the URL you can retrieve all the summary information then the important information for this particular PDB code from the PDB database. So the idea is to generate a collection of endpoints like these ones that you can see here from all the different building blocks integrated in the bioxial building blocks library, which are a lot. And for that, we decided to go in an automatic way automatic process to generate all of these code that is behind these endpoints but we actually succeed on that so we had all the we have now all the different building blocks available as the rest API endpoints. The implementation is a bit complicated but you can see here a schema of that that this is the user this is the base URL for the rest API the user contacts through this URL to our machines in our premises and these machines are connected to a database. This is exactly the same as a typical rest API, which retrieves information from a database and gives this information back to the user. In these cases a bit more complicated because we need to run the tools that are behind these bioxial building blocks and for that we have a private cloud infrastructure in our own premises in the MMB group inside the institute for research in biomedicine in Barcelona, all these machines. They have all the bioxial building blocks installed using conda environments, so that when a user uses or calls one of these rest APIs. The functionality the tool is running is executed in this cloud infrastructure and the result is given back is safe to the database and given back to the user. This is the process in time, because this is most in most of the times in our executions are asynchronous they are not synchronous because they need time something. If you think about an MD simulation, it needs some time hours, even so the user sense the input and configuration files to a rest API. The job starts in our virtual machines and a job ID is returned to the user so this token ID is then used to pull that means that the job status is checked every time. And then once the job is finished, the output data is saved in our database and the idea of the output is returned to the user. Then with these outputs, the user requests the outputs and the files are retrieved from the database as in a typical rest API, and the user can download the output files. It seems really complicated, but it's not so complicated to use and now I'm going to show you how to use these rest API web services. You have a web server for this and here is the URL with all the information that you need and basically information. It's very easy, sorry, it's very easy because you have a couple of tabs here the availability and the tutorials I will go through them. In the availability you have the genetic endpoints genetic that means that, for example, you can get a list of all available packages those are endpoints right as I was explaining you before. You can get a list of all the available tools for a given package and when I say package here is a module in BIOXO building blocks, for example, by we be able to skinning by we be grown max you put here the name and you get a list of all the available tools. If inside the package you put a tool a particular tool you can get information for a given tool and if you do that in post you can launch the job and I will give you now examples of how to do that. And these are the retrieval of information from the outputs generated by this rest API endpoints with the token that is received when you launch the building block or the rest API. And then you can also retrieve sample files in case you want to see how are the files that you need to use as an input for the rest API. We have the tools endpoints that means all the different endpoints available in the rest API, the categorized again in different modules so for example chemistry and these are all the different endpoints for the chemistry tool. All of these can be found in the web server and this is this is using swagger with the open API standard if you are familiar with this that means that you basically in the website you can test and try the this rest API as you can also use your own input. And use your own input to try and execute this rest API is through the website. And finally, and I think this is very nice, we have a tools execution section in the in the website that lets you try in a very easy way all these collection of endpoints. This part of the website includes four different steps. The first one is the selection of the tool so the building block and I will do all the different steps. The four different steps with a particular example which is finding pockets in a protein using the f pocket software that is very, very famous. So we select the biaxial build a virtual skinning package and we select the f pocket run tool. We know that this is the one that we need. And then we use we put as an input a JSON file with a configuration file now this is the only if you may be complicated the step because you need to know what is inside this Jason but you have here a link to the f pocket run tool which is giving you all the information that you need to produce this Jason file. And if not, you can also download the sample file here and just modify the sample file and use the sample file as an input so very easy. And then a path to the PDD structure in your own computer, and then a couple of names for the output files, and this is going to do this part of the process of the time so it's sending the input data calling the rest API, the data will start and a token ID will be received, but you don't need to do anything about with this token ID because the web server is doing that for you. So, this is just an example of a Jason file for this particular building block in the website is going to do the polling for you remember this is the polling part of the process. So, the website is checking the job status and is giving you the information about okay, we still need to wait because the job is being run in the virtual machines when the job is finished then it will receive the ideas of the different outputs and the website is going to tell you okay the state is finished and these are the output files. Again, there's a couple of IDs for this or token ID for these output files and the website is automatically generating these endpoints to retrieve the output files that are generated by the rest API so all of this can be used can be done using the website the rest API website so I encourage you to take a look at the website and this sorry this is the final part of the process which is retrieving the output files from the database. Okay. Apart from that and of course I'm interested mostly on workflows you can also build workflows using this rest API. And this is an example of a Jupyter notebook on how to use this programmatic interface, and it's very easy we have produced a bit of a library or a library for that that is just launching the job with a particular rest API endpoint with again the configuration for the for the endpoint for the building block and the output in this case which is basically we are downloading a PDB from the PDB file checking the job is doing the poll and retrieving the data once the job is finished and then you can visualize the information this downloaded PDB file in Jupyter notebook. You can do something a bit more complicated like the example that I was showing you before about finding pockets in a protein. And this is the way and you see it's exactly the same you generate the properties or the JSON file the configuration file. This is the endpoint that you want to run. This is the outputs where you want your data to be placed. You check that your launcher job you check the job and you retrieve the data so in all the different building blocks from all the collection of building blocks in the BIOS building blocks library. You can run them like this in the same syntax for all of them. And remember that this is run in not in your computer but in a remote computer. And again, if you are interested in workflows as I am you have examples in the website about how to run an MD set up again to prepare the protein to run a more dynamic simulation using the rest API so everything in a remote way. So you can go to the website and take a look at this, which basically have all of these different steps. You can go from fetching the PDB structure, adding feeling so creating the solvent box, adding a water molecules the ions minimizing the system equilibrating the system and also running a short free under strain molecular dynamic simulation. The only thing that you need in your machine to run this workflow is basically a Jupyter notebook with Python of course and the possibility for the Jupyter notebook to read the conda environments. If you see intermediate results, these libraries about 2D plots, 3D representation or trajectory representation but this is the only thing that you need to run this workflow and the installation processes. You can find the installation process in the documentation for the for the workflow. Okay, enough about the rest API and the programmatic interface and now jumping to the next part of the of the presentation today which is the bio BB workflows. In this case, and as a difference from the previous one, this is a web based totally totally web based graphical access to biomolecular simulation workflows. Now we are focused on workflows and we are focused on the web based graphical axis not just to possibility to access remotely to them by accident blocks, but also to the workflows. They are still running the providers infrastructure in our case in this case in MNB in our group, of course again with limitations. But as you will see in the example that I will give you the workflows can also be downloaded and run locally in your own machine in case that you want to tune the workflow modify some of the steps or modify the parameters. This is a personal workspace with all the users projects and I will show you something about that and this is very important. And again there's no need of an installation and or deployment in your machine. In case you want to run them in the website and not locally. There's also the possibility to connect the web server to external clusters I will tell you something about this in the last slide of the presentation. The implementation of this web server is a bit more complicated than the rest API before but the idea is more or less the same actually the first part of the implementation is shared with the rest API. That means that we have a database with all the information we have the private infrastructure in our, in our own premises. With all the installation already there all the biops of building blocks there. And then we have a typical file system this where the trajectories are written. And then the trajectories can be streamed directly to the user using this fantastic MDSERF technology that if you are not familiar with it just take a look at the publication in NAR. Last year, because I think it's fantastic it's basically streaming all the information from the trajectories to the user through website. If you remember the collection of transversal workflows that I was telling you before from the biops of building blocks. There's also a possibility to run to launch this collection this this workflows, using this by actual workflows. A website. Most of the demonstration workflows that we have for the biops of building blocks are integrated in the bios of workflows in this tool that I'm going to present you now so basically set up of a protein or a DNA or or a protein with legal using number using Gromax with the possibility to mutate a residue or more than one parameterize a legal run a docking with a protein and a legal analyze the trajectories from MD simulation and also a bit of DNA analysis DNA specific analysis all of all of these workflows are integrated in the in the collection of workflows website. And how to use this is very easy, but say easier than the rest API in this case you can connect to a web server and start from either from workflows part or from a creation of the project. If you start from workflows you can just select which of the workflows integrated in the website is the one that you are interested in. And if you want to create a project from a structure you can go here from structure if you have a sequence and you want to model the DNA or RNA structure from a sequence you can start from here. If you want to analyze a trajectory if you want to run a docking or you want to parameterize it from small molecule you can start from here the reason that we have divided this is because, as you can see the input is completely different so the website is customized to this type of input, so you can use some examples of that. Starting from a structure, for example, those are the different steps in the website. Starting with the provision of a particular structure you can upload the pdb file from your own computer you can use a pdb ID, you can go to the alpha fold database you can even browse for whatever you want in the pdb database in this case I'm looking for pdb kinase for example and if you go here, sort of here I'm just highlighting you that there's different workflows that you can select for this particular input so for this particular structure, you can select to set up an MD simulation with Gromax to set up an MD simulation with a protein and a ligand with Gromax the same with amber and also with Gromax and mutations. If you go to the part of a kinase and you click on search, you will find and as you can see it's very graphical table with all the different part of a kinase from the pdb and then you can click. You can also take a look at the structure here but you can click on one of the pdb and use this one as a structure to start the workflow the execution of the workflow. But before starting the execution of the workflow, we have different sub steps inside the web server the second one is to check the structure here which I think is a very important process. For a molecular dynamic simulation that most of the times we forget about that and I think is very important. So for example this section of the website is giving you information about okay you have some. At the atoms here take a look at them. Those are the sort of take a look and see and think if they are important for your simulation or not. Incorrect amide assignments like for example these two nitrogen atoms here that are very close to each other that basically is just a way that you need to solve this problem you just need to swap this. Nitrogen this oxygen because sometimes when you solve the structure by crystal or crystallography you have this problem that you cannot identify. If this is a nitrogen or this is an oxygen this kind of things happens in every pdb so it's good to take a look at it and also backbone breaks for example. And as you can see here there's a big backbone break that can modify completely the flexibility and dynamics of your simulation if you start with this structure so I think that this step is very important and the web server is. Giving you all the important information that you need to be for launching the simulation or the preparation of simulation. Is to modify the settings of the workflow settings are reduced in this case. You have not all the different settings that you can modify your parameters that you can modify we we have done this on purpose because it can be a little bit complicated if you have all the different parameters here and I will tell you. The way to solve that here you can change the workflow so when you are at this step you need to change you need to think about what is the workflow that I want to use. You have all the information about the different steps of the workflow. This is exactly the same that I was showing you before with 14 different steps generating the. Just downloading not downloading using the structure that you do you have used as an input creating the topology solvent box ions minimization acceleration etc. You have this information for all the workflows of course, and then you can also modify. The parameters and the final and the configuration files that means that basically the database web server. Is able to generate for you the configuration files for a long and the simulation that you can then download and use in your own infrastructure to run this case, for example, I think that was a 10 nanosecond simulation in your own machine. So the setup is going to be running the server the last the final and the simulation you have all the configuration files that you need to run them in your own machine. And this part is important because in the third step there is also a kind of a summary. Which is, if you want a provenance of all the of all the project, the name, the input structure all the modifications that you've done in the checking all the parameters that you've written or modified in the previous step. And here is where you can also download the workflows to run them at home and there's information to different flavors Python pure Python workflow CW welcome on workflow language workflow. To different ways, very easy to reproduce all the information is there to generate the condom environment with all the dependencies remember that these are by actual building blocks so you can build an environment to run the workflow in your own machine. And also you can just see WL that is using Docker containers and in this case you don't even have to bother about an environment that Docker containers will be downloaded. The only thing that you need is a Docker engine of course installed in your machine. So you can tune the workflows if you want downloading them and running in your own machine if you don't want to run them in your own machine you want to use the server then you click on next, and the workflow will be cute in our system in our queue system in our private cloud infrastructure and will be run, and you have all the steps of the different workflows. And finally you will receive the final page with the basically the analysis results which are the ones that you are interested in the analysis results again are customized depending on the workflow so he disease the analysis of MD setup workflow with the initial structure with the final structure this is the interest train the trajectory for the interest train. MD, last step of the workflow, and here energies radius of generation and RMSD during this interest train simulation the last step 500 picoseconds. So if you think about protein league and docking you have different outputs like this one where you can select the different poses here is auto dock different poses and you can see the pose of the league and inside the pocket of the protein. This is an analysis of an MD trajectory and this is an analysis of DNA specific and the trajectory with this case the slight helical parameters and the correlation for different base pairs. Each one of the workflows have different customized output results, and all of these you can find information in the health page of the web server. Now, if you go to the FAQ of the web server there's one question that is very important, and is that is strictly necessary to be registered in the baby workflows you can register as you can see here and have a user account. And I strongly recommend you to do that if you want to try the web server. Why, because it's very easy there's nothing else than this just entering the name and the email and your institution. And then, when you go to the project actions of the results of a workflow, instead of having just downloading the results. All of these different options which are rerun the project clone the project make the project in persistence so that did not disappears after one month and also make the project public so in case you want to share all the project the URL can be exported and give them to a colleague. And all of this information available for them, which at the same time they can also take these and rerun a project or clone the project in their own user account so very interesting options. This is the user workspace with all the different projects that the user has run with all the different actions that you can do this five gigabytes for each user. You can see from the user profile you can see here the different profile you can modify all the different information that you have used to generate your user but these here it will give me the opportunity to talk about these SSH keys which is the last slide for this project which basically is here, and we also give the possibility through these SSH key to connect to remote HPC clusters and we have tested that with different supercomputers in the personal supercomputing center and also in CSC and Pugti machine. So this is the computer this is the user you're of course your account. This is a folder in the in the supercomputing machine and this is an SSH key that is generated automatically by the server when you click on this button, and the check is checking if the server has connection using the user and the SSH key to the computer. And then once once it has connection, you can run molecular dynamic simulation, long molecular dynamic simulation using this connection and then the files will be sent to the HPC, the SLURM or whatever job will be launched there, and the output will be retrieved and all of these also thanks to another module of the VIOSO building blocks which is called BioBB Remote. Okay, so I'm going to finish the presentation with some conclusions because I think it's time. As a summary, I have presented in a very quick way a couple of projects from the BIOSO building blocks ecosystem. The first one is the collection of fresh API endpoints that give us remote and problematic access to all the different building blocks integrated in the BIOSO building blocks library with no need of installation and running in the providers infrastructure. And on the other hand, I have presented the graphical the web based a graphical interface to access this set collection of demonstration biomolecular simulation workflows that are again running the providers infrastructure but it also give you the possibility to download them and run them locally with a personal workspace with all the users projects that can then be shared to your colleagues, no need of installation and connection possibility to connect to external clusters. Let me finish please with a knowledge and acknowledgments to the people that basically run all the work which is Janice, our full stack developer, Professor Modesto Rothko is the PI of the group. And in the Barcelona Supercomputing Center we have the help of Professor Giuseppe Giuseppe, and also Pao Andrea which is the main developer of the BIOSO building blocks. And finally, we have time now for questions or suggestions. Here you have a two publication for the two different projects, please take a look if you are interested. And I have put here a bit of a questions for you to start thinking about that. It can be now or it can be after the presentation after the webinar if you will, if you are interested in playing a bit with the servers or reading the papers, please then contact me and give me your feedback, sorry about what functionality or workflow would you like to see integrated in these tools. This is important because in BioXL we care about our users so we will take a look at this feedback and we will try to integrate if possible what you think is important for you. I would consider using them these tools only if something needs to be changed, something that you can think, please tell us and find them useful or not useful because please tell us the reasons. So thanks for being here and now it's time for questions. Thank you very much Adam. Yeah. So, while people are starting to type a question so we got a question from Patricia. She asked the question that I have is how the workflow helps trainees to become better simulators users. Okay, I think that for the trainees it makes you aware and this happened to me all the time that first you need to, you know, it's maybe it's not in the workflow part but it's in the checking part that I was showing you in the workflow server that when you start the simulation you need to take a bit of time to take a look at the structure and identify possible problems. I mean backbone breaks that I was showing you something related to the important legans, important protonation states of different residues. This is very, very important and it's not even in the workflow by itself it's before the workflow, and you need to be very aware of all these possible problems because if you start with something that is not correct. Then, the, as I was telling you during the presentation you will end up having a dynamic or a flexibility result that are completely wrong. So, sometimes for the trainers also, we need to be aware of that and when we start working with these automating processes and we start running 100 different and these relations at the same time with 100 different structures. We need to be aware of that. This is one of the points. I can think about more points but I think that maybe we can discuss different questions today. Maybe it's also good for a building tutorial. What do you mean? What do you think about Adam? Of course, of course. So the demonstration workflows, I didn't have time to enter into that in detail but for the demonstration workflows that we have in the in the BIOXEL building blocks website but also in the workflows website in both of them. They are transversal workflows that means that are workflows that we are using in every day in our everyday life. So the preparation of the setup something like this. And we then can use these transversal workflows in our training events and actually in BIOXEL we have been using these workflows since I think it's been like four years that we are using these workflows. We are developing more of these kind of workflows. I think that they are very interesting and in particular the Jupyter notebooks one. Because there you have all the intermediate results and you can inspect the intermediate results and if something is not really as you wanted you can go back one cell. I hope that you're familiar with Jupyter notebooks, modify the parameters of this cell and then run it again take a look at the intermediate result and go that way until you find the whole workflow with all the parameters that you are interested in. So yes, absolutely. While people are thinking and maybe are typing a new question because please use the Q&A function to type your question. In the meantime, I have another question for Adam. What do you think is the higher barrier or better. What people need as a background to start to use the building blocks. The API or the workflow that you show it. I recommend to start with the baby workflows. So the last project the last tool that I have presented today because basically is a graphical user interface, and it's very easy to use. You start with a pity code for example and try to run this MD setup workflow from a structure. Once you have run that and look at the different steps and identify what the different steps are doing then you can jump to the if you are interested you can jump to the BIOXO building blocks website, the main one where we explain all the building blocks and the library. And look at one of these demonstration workflows which are you with the notebooks and each of the different cells of the notebooks is a different building blocks or a different step of the workflow and it contains information about what this step is doing. So, let's see that the BIOXO workflows website is giving you a kind of a black box with the whole workflow and you can run the whole workflow. And if you want to start thinking about what are the different steps of the workflow and what are they doing, how can I modify this, you can jump to the Jupyter notebook and then modify these parameters. And then we can go on and on, and we can end up doing massive things in HPC machines but there's no time for that for today. Many people need a base of Python, a base of Jupyter notebook, not to allow to use Jupyter notebook, you, Linux, Shell, what do you think is the minimum? I don't think they need much because if you open a Jupyter notebook, one of these demonstration workflows and you can just run one click in the run button to each of the cells you can directly identify how to modify the parameters because it's very easy you have them. You don't even need to know a lot of Python, you just identify where are the parameters and just modify them. So, yes, if you know Python or you are familiar with Jupyter notebook that is, it would be easier for you, but for me it's not a limiting step. Okay, thank you. Please do that in this, if you can, if you want to ask questions also you can ask whatever, you can ask also what are the future idea that where in which direction they will go this workflow, you can ask whatever you want. In the meantime, people are thinking I just go through some announcement, so people can type while I make some announcement. So, by Excel has different activities so here, so I want to go through a little, this is the new webinar that we will run in June. So we will have a webinar on a realtor research environment for integrate modeling of bi-molecular complex with the new modular version of ADOC that will be given by the Sonderburg van from the University of Utrecht. And this will be the 30th of June, and then one week after, we will have another webinar on another course software of bi-excel, this PMX, and that is Gromax PMX for an accurate estimation of free energy difference, and then it will be given by Sudra Mashabea from the Max Planck Institute of Göttingen. Then we all, we have also summer school where all you will learn how to use the building blocks, also Gromax and PMX, they are also topic of this school and this will be in Sardinia from the 10th and the 15th of September. So, you can still register so please go to the website of the bi-excel if you are interested. And finally, bi-excel is a running for all the three code. So the four software cores, so that means Gromax, PMX, the building, so the library, the building blocks library, and ADOC, that we will run a survey that we will use to build our plan for the future. And then we want to understand, like Adam was mentioned, the direction that those softwares to go to fulfill user need. So now if you go to here, one of the survey that is already online is the Gromax survey. So I will, if people are using Gromax, please also if you're not using and you want to give your feedback, please go there and make the survey. And soon we will have also the survey for BioBB building blocks for PMX and for ADOC, and probably Adam will post the link to the survey in the forum that as I was calling is called ask.bioxcel.eu, so you are all welcome to go there to ask a question to Adam also after the webinar. So, please go ahead with some other question, just type it in the Q&A panel, if you want. Patricia did Adam answer correctly to your question, or did you have something, some other reflection? Yeah, Patricia say thanks. Okay. No other question are coming. I thank you all. Thank you very much for attending the webinar. Thank you very much Adam for giving this webinar. And please go and try to use the BioBB building blocks. If you make mistake is not a problem. You can just ask Adam and the others so just it doesn't happen nothing I think you don't broke anything. Okay. Thank you very much for attending. Bye bye.