 Hello. So I would like to thank the organizer for giving me the opportunity to present the NFCOR community at Jobim. That's a very nice opportunity for me. And so let's begin. So I'm working at the Swedish childhood tumor biobank, which is located within the Karolinska Institute here in Stockholm. And I'm also sitting half time at the National Genomic Infrastructure, which is a sequencing facility used by researcher all over Sweden. And I'm here at the time to develop pipeline, because we do need analysis pipeline also for the biobank. And we decided to be together and to share the effort in developing it. NGI is part of CIDAFLAB, which is a national center for molecular biosomes. And CIDAFLAB is comprised of several infrastructures. And with NGI, we do also collaborate a lot with the National Bioinformatics Infrastructure of Sweden, which is NBIS, which is the local election node. For me, as a bioinformatician, I think that reproducibility is a central matter. And yes, and I think that's why we really like started to use the next flow. So for us at NGI, we were early adopters of the next flow for its portability, for its shareability, and of course for the reproducibility. Next flow is data driven language, which means that the way to link like all the different tools within the pipeline. So the execution graph depends on the input data and it's calculated on the go. In state make, it's actually the other way around. The execution graph depends on the final target and it's calculated before launch. And for me, that's the main difference between SnakeNake and next flow. Just the logics are different. But apart from that, I do think the language do provide similar possibilities and I'm just happy as long as you use a workflow manager. So next flow supports the main scheduler on regular HPCs and also on the cloud platform for reproducibility. It supports virtual environments such as Conda and also container engines and docker and singularity. Let's talk about NFCore. So at NGI, we started, we have been developing analysis pipeline for years and we have been using a set of standards that helped other group run the pipeline on their own. And the pipeline began like to outgrow the NGI and the silaflap branding. So in late 2017, Phil Jules from NGI, Alex Pelzer and Sven Finger from QBIC and Andres Vim from ASTAR decided to merge all their effort together. And we created the NFCore organization and moved all the NGI pipeline in this new organization and we removed all the bonding and everything. And then we started working also on new pipelines. One year later, we were 12 institutes. One year last year we were 27 institutes and nowadays we are 39 institutes. So we have a website, you can use this link to have more information about how to join the community. We are active on Twitter. We also are using YouTube more and more especially nowadays with the pandemic and the lockdown. We do have a slack as well and I think I might be using that a bit too much also. And of course we have GitHub. GitHub is where we organize all our different propositions for the pipeline and for the website and the tools that we provide as well. On our website, we also released all the pipeline that we provide. So currently we have 25 release pipeline, 14 under development. This is the pipeline that we have been released most recently. So we worked on this viral recon pipeline was led by Archille Patel from the Francis Kitt Institute. And it's about analyzing data from the COVID-19. This is my own pipeline but to detect germline and somatic variant from world genome sequencing, exam sequencing or targeted sequencing. We also have a pipeline for mass cytometry analysis, SLAM-SIC, corporate host identification and also a pipeline to identify and quantify peptides from mass spectrometry data. Here are our most starred pipeline. So we have an RNA-SIC, chip-SIC, mesil-SIC, anti-SIC, attack-SIC pipeline. Of course, SARAC. I like my pipeline. We also have a pipeline to detect gene fusion and a pipeline to analyze ancient DNA. For a pipeline to be able to be an NFCOR pipeline, it has to be a next-flow base pipeline because we are strongly tied to the next-flow community. We all choose to use next-flow and we want our pipeline to have the same common structure which is based on a template that we provide. They need to have a stable release tag that helps us for reproducibility, of course. We need to have MIT license, that way it cannot be used in commercial settings. We do strongly advise the software to be bundled for reproducibility and we do also advise to use continuous integration testing. Apart from pipelines, we do also provide a companion tool, which is called NFCOR tools, that helps with the most common tasks that we can do. So you can use it to release all the available pipelines within NFCOR. I run a pipeline with interactive prompts asking you about the different parameters that need to be selected. You can also download a pipeline for offline usage and then transfer the pipeline, the code and the container on a server which is completely cut from the internet. This is our case in Sweden. We have an offline server for human data. You can also list the licenses that all the tools in the pipeline are using. And of course, and I think this is important for us, you can create a new pipeline from the template and check pipeline code against guidelines. So to create a new pipeline, it's actually quite simple. You just type NFCOR create and then you just need to type the name of the pipeline, short description of the pipeline, your author name. And this will create a skeleton, a minimal skeleton that does provide, at least I think, one multi-QC process for reporting. And we are trying to, we are at the moment, we are working to make it less specific to NFCOR to also help the whole next-flow community. Also, when we are making a pipeline, we do need to take care about software dependencies. We are quite involved within the BioConda community and we do maintain and add recipes from time to time. And we do advise that the way we do that in NFCOR is that we install all the tools if possible within a single Conda environment that is actually bundled into a Docker container. So next-flow, that way you can download from Docker either as a single IT image or as a Docker container. And that setting allows us for an easy update of the tool because we just have to change the version number in the Conda environment file and easy usage of all the different containers, engine or virtual environments that you want to use in your system. One thing that is also very important for me is the configuration. We do try to provide all pipeline with default sensible configuration for a regular size cluster. But of course, you need to specify which scheduler you are using and stuff like that. So to help people from the same institute sharing pipeline, we do provide a GitHub repository that is being checked when you launch a pipeline. That do allow to share configuration between pipeline for a specific HPC. So it means that if someone in an institute managed to make an NFCOR pipeline work by specifying the right amount of CPU, time and memory environment, the right scheduler, the queue if available or the environment, the path to the common reference file, et cetera, et cetera. Anyone else from the same institute will be able to run any other NFCOR pipeline. So as I said earlier, we are quite involved with the next developer and we are up to date with the latest development. At the moment, we are working on making full size dataset testing for pipeline releases on AWS. We are also working to make a graphical user interface to launch pipelines. And with the latest version of Nexlo DSL2, we are developing modules which will allow for more modular pipelines, which will be similar to the Snakemake rule files. Let me present you the core team. So these are the people that do help in administrating the NFCOR community. If you are interested and you want to join the core team, it's possible we are open for a position. And so UNS, Phil and I, we are from Stockholm. Alex, Sven and Gisela are from Germany. Archive is in the UK and Olga is in the US. We also have a correspondence in natural biotechnology. Also, yes, Phil is a little bit crazy about statistics. So we have extensive statistics about anything that you would think of on the website. For example, we have a contributor leaderboard about every contribution that has been made on the master bunch of all the repository within NFCOR. You also have information about the minimum response time for issues or food requests and so on. Of course, we're a community, so we do also some events. That was the last hackathon that we made in London at the Quik Institute. That was just before the pandemic. We were planning to do one hackathon in Germany quite soon. And it will still happen, but it will be a virtual one. It's still open. So if you want to register to it, feel free to it. There will be some tutorial to help you make your own pipeline to help you collaborate on the pipeline and also to help you use pipeline and such. And we will also try to work on other core features that we want to be implemented. Yes, so as a stay-at-home message, so yes, I just want to recapitulate what I said. So we do provide a pipeline with good reporting for facilities with validated release so that we have reproducibility. For the user, we have portable and easy-to-use pipelines that are well-documented. They are easy to share between different collaborators. And for the developer, we do provide templates and tools to help you simplify the common task. I would thank the organizer for giving me the opportunity to talk at this session. I would like to thank all the institute I'm working with or at. I would like to thank all the institute within the NFCore community. I would like also to thank all the contributors for the NFCore project. And of course, I would like to thank all of you for listening to this talk. If you have any questions, please feel free to ask. No. Merci beaucoup Maxime. So there is two questions on the Slack and on the Slack, but the Q&A panel. So the first one is from Jean-Diane Corrier. So do you plan to build links with the EOS Clive project Workflow Hub that is developed by European Bioinformatics Institute, by the Carol Goebbels group, for example? I'm afraid I'm not aware of this initiative. So I will look more, but I think we do want to collaborate with more people. The whole point of making this project was to avoid developing the same stuff on the separate side. We do want to collaborate if it's possible. Gilles, you want to say something? It's okay. So it's just a registry. It's not something too... And it's agnostic. So yes, there was a repropose on SnakeMix, on Next Close, on CWVL and Galaxy Moreflow. It's a duplicate effort. Yes. You can turn it. So it's Workflow Hub.eu. Okay, no, no, but then I will look more. I am pretty sure some other people within NFCore are aware of that, but yes. So we have a second question. How do you test the reproducibility of your pipeline to ensure their work in a new platform? So for us at the moment, what we are doing, we do provide containers that contain the same version number and that are tagged for a specific version. And if you use that, you should be able to have the same reproducibility. Next floor, I think if I remember well the paper, they did try that and they proved that it was reproducible. So for us, that's enough. I think that's what we plan to do with being able to make the life-size test within AWS. And I think at some point also I would like that we could be able to develop some more tests. At the moment, we are just doing some continuous integration testing. But I would love to make some validation testing if it's possible using a golden set of data tests or stuff like that. But it's kind of difficult because we do provide pipeline for different kind of data. So I think it's something that needs to be done on a pipeline basis.