The de.NBI / ELIXIR-DE training platform - Bioinformatics training in Germany and across Europe within ELIXIR

The German Network for Bioinformatics Infrastructure (de.NBI) is a national and academic infrastructure funded by the German Federal Ministry of Education and Research (BMBF). The de.NBI provides (i) service, (ii) training, and (iii) cloud computing to users in life sciences research and biomedicine in Germany and Europe and (iv) fosters the cooperation of the German bioinformatics community with international network structures. The de.NBI members also run the German node (ELIXIR-DE) within the European ELIXIR infrastructure. The de.NBI / ELIXIR-DE training platform, also known as special interest group 3 (SIG 3) ‘Training & Education’, coordinates the bioinformatics training of de.NBI and the German ELIXIR node. The network provides a high-quality, coherent, timely, and impactful training program across its eight service centers. Life scientists learn how to handle and analyze biological big data more effectively by applying tools, standards and compute services provided by de.NBI. Since 2015, more than 300 training courses were carried out with about 6,000 participants and these courses received recommendation rates of almost 90% (status as of July 2020). In addition to face-to-face training courses, online training was introduced on the de.NBI website in 2016 and guidelines for the preparation of e-learning material were established in 2018. In 2016, ELIXIR-DE joined the ELIXIR training platform. Here, the de.NBI / ELIXIR-DE training platform collaborates with ELIXIR in training activities, advertising training courses via TeSS and discussions on the exchange of data for training events essential for quality assessment on both the technical and administrative levels. The de.NBI training program trained thousands of scientists from Germany and beyond in many different areas of bioinformatics.


Introduction
In the last decade, researchers in the life sciences have been early victims of the 'Big Data Problem' because of technical improvements in the so-called 'omics' and image analysis fields including the challenges of the five 'V's of big data: volume, veracity, velocity, variety, and value 1 . Here, large and complex datasets are rapidly generated to analyze various biological levels in living cells per day. Even small benchtop sequencing machines are now capable of producing terabytes of data, but many life scientists neither possess the skills to analyze the data properly nor the knowledge to use existing analysis resources. Therefore, the true bottleneck of the current 'Big Data Problem' in the life sciences is often not the storage or compute power, but the knowledge and skills how to use existing bioinformatics services and tools. An important way to solve these deficiencies in the field of life sciences is by means of bioinformatics training. The need for such training was described recently: The majority (> 95%) of life scientists located in Europe work or plan to work with large datasets, but less than 35% possess the bioinformatics and statistical skills to handle the huge amount of generated data 2 .
Introduction to the German Network for Bioinformatics Infrastructure (de.NBI) In 2012, the German Bioeconomy Council published the recommendations 'Requirements for a Bioinformatics Infrastructure in Germany for future Research with bioeconomic Relevance', coming to the conclusion that expertise centers in Germany should have permanent structures for training users from life sciences 3 .

Introduction to ELIXIR-DE
Since August 2016, Germany is member of the European lifesciences infrastructure for biological information, ELIXIR. The Federal Ministry of Education and Research (BMBF) signed the ELIXIR Consortium Agreement establishing Germany as member no. 20 of the distributed infrastructure that unites Europe's leading life science organizations. The German Network for Bioinformatics Infrastructure (de.NBI) has been designated by the BMBF to establish the German node in ELIXIR by providing high quality bioinformatics services and training expertise to the scientific user community in Europe.
On 9 January 2020, the National Collaboration Agreement to establish the German ELIXIR Node (ELIXIR-DE) became effective. This contract was signed by 21 de.NBI members (German universities and research institutes), including Bielefeld University that will take over the legal role as 'Representing Entity' of ELIXIR Germany. The National Collaboration Agreement is the legal document that specifies the organization of ELIXIR-DE and the role of the Central Coordination Unit (CCU).
In addition, the ELIXIR Collaboration Agreement was signed by the Representing Entity and the ELIXIR Director, Niklas Blomberg. This contract became effective on 10 February 2020, thereby fully establishing the German ELIXIR Node. The ELIXIR Collaboration Agreement is the legal document that ties together the ELIXIR Hub with the national node and it forms the legal basis of Commissioned Services. These technical projects guide future service developments.
Although not being part of the German ELIXIR Node, EMBL Heidelberg is a consortium partner in de.NBI and is a strong collaboration partner and contributor of the German ELIXIR Node through its own ELIXIR Node (Node to Node collaboration). The list of services provided by EMBL to the German Node has been included in the EMBL ELIXIR Work Programme, which has been approved by the ELIXIR Board in April 2019. The EMBL ELIXIR Work Programme therefore represents the legal basis of Commissioned Services carried out at EMBL.

Introduction to the de.NBI / ELIXIR-DE training platform
The special interest group 3 (SIG 3) 'Training & Education' is centrally coordinating the training courses and the education efforts within de.NBI and ELIXIR-DE. It is composed of

Amendments from Version 1
The text was reworked based on the reviewer comments. The introduction includes new chapters about de.NBI, ELIXIR-DE and the de.NBI / ELIXIR-DE training platform to give better insights into this term and show the difference between de.NBI and ELIXIR-DE. Especially, the service center text was written more concisely and was more standardized than in the old version. In addition, we made an update of the training data including the new data of 2019. In total, 3 figures were deleted based on reviewer comments and a new table and a new figure were included to fulfill the reviewer wishes.
Any further responses from the reviewers can be found at the end of the article REVISED 19 training experts from each service center and is led by Oliver Kohlbacher (Eberhard Karls University of Tübingen -Chair) and Daniel Wibberg (Bielefeld University -Deputy Chair) (status as of July 2020). SIG 3 collects all planned de.NBI training activities and provides a structured process of reporting, documenting, and monitoring all training events. The general goal of de.NBI is to develop a national bioinformatics training program for Germany.
The following topics are currently in the focus of SIG 3: • Improvement of the quantity and quality of training courses.
• Introduction of online training/e-learning courses/ platforms.
• Standardization of training course monitoring.
• Integration into the ELIXIR training platform and TeSS.
• Coordination of training events and standards with ELIXIR.
The strategic planning of SIG 3 includes following aspects: • Closing of gaps in the topics covered by the training courses.
• Improvement of the quantity and quality of e-learning material.
• Qualification of more trainers, e.g. by more Train-the-Trainer courses and Trainer meeting.
The training activities of each de.NBI service center are described in the following sections.  2016, a partner project on computational epigenetics (de.NBI-Epi) has been added to HD-HuB, which is coordinated by the researchers of Saarland University in association with DKFZ.
Research focus. HD-HuB service center focuses on delivering computational resources to support the development and maintenance of bioinformatics tools in four main application areas within the organizational framework of de.NBI: 1. High-throughput sequencing (HTS) data analysis to study human genetics and genomics 6 .
2. Taxonomic and functional microbiome profiling, genome and metagenome annotation and association of microbiome composition with host phenotypes to facilitate metagenomic studies 7 .
3. Automated analysis and visualization of large-scale high-content phenotype screening data for phenotyping of human cells 8,9 .
4. Analysis workflows for genome wide DNA methylation data for computational epigenetics research 10 .
In addition, HD-HuB members established two of the six de.NBI Cloud sites in Heidelberg and Berlin.
Training and outreach activities. The HD-HuB consortium has offered training courses addressing the computational needs of early stage and advanced researchers with a wide variety of research interests. The de.NBI Cloud compute resources are used in different HD-HuB training activities. Below, HD-HuB's training activities are listed under different categories based on the different learner-profiles: 1. Programming courses for novice and advanced learners. HD-HuB members from EMBL organize several beginner and advanced courses on programming skills every year in Python, R, Unix/Shell and version control. These courses have been positively received by the course participants, who constitute ~50% of our overall learners every year. Such courses have been vital to the computational skill development of biologists and wet-lab scientists, who lack formal training in computation and bioinformatics. In addition, The Carpentries courses have been very popular among the novice learners in our communities that allow them to acquire the basic building blocks of software development at the Software Carpentry workshops and get hands-on experience of data analysis at the Data Carpentry workshops.

Training on specialized bioinformatics topics.
In addition to offering general training courses, the HD-HuB service center offers specialized courses as per demand from the researchers with different research backgrounds. These so far include courses on protein bioinformatics, bioimage analysis, DNA methylation, experimental data analysis and RNA-Seq and Bisulfite sequencing data analysis in cancer research.

Courses for advanced computational biologists.
To support computational needs of advanced learners in our community, we offer hands-on training on data analysis skills such as machine learning, workflow management, cloud computing and statistics. These workshops also provide an opportunity for peer-based learning by bringing scientists with similar research interests together who can potentially collaborate with each other in the future.
4. Train the trainer courses. Trainers at our workshops are often volunteers from the respective scientific community, who spend time in creating useful resources to teach at these events. These volunteers are highly valuable to our mission of training computational skills to the future learners. In order to facilitate such knowledge transfer effectively, HD-HuB has started to train experienced researchers, who can become trainers in future courses at their respective institutes. These trainings are offered under the title 'train the trainer' where they can gain theoretical and practical knowledge of useful teaching techniques used in best teaching practices.
5. Local outreach activities. HD-HuB hosted the first international de.NBI symposium 'Bioinformatics for Human Health and Disease' in 2016 at the German Cancer Research Center (DKFZ) in Heidelberg. Attended by 140 participants and over 20 international speakers, the conference provided an opportunity for the researchers to discuss their research and plans within the de.NBI framework. HD-HuB has also co-hosted local event series like de.NBier technical seminars and the Heidelberg Unseminar in Bioinformatics, which are mainly targeted at non-expert scientific communities.

Bielefeld-Gießen Resource Center for Microbial Bioinformatics (BiGi)
BiGi consists of three partners: the Genome Informatics group at Bielefeld University, the Bioinformatics & Systems Biology group at Justus-Liebig-University Gießen and the partner project at Otto-von-Guericke-University Magdeburg.
Research focus. The BiGi service center offers tools, services and training for microbial genome 11-14 , metagenome 15 , and postgenome research 16 that is complemented by a large-scale hardware infrastructure. The focus in Gießen is on genomics (assembly, annotation and comparative analysis and shortread-mapping data evaluation) and read-based metagenomics research. Bielefeld focuses on assembly-based metagenomics and post-genome analysis, such as transcriptomics, proteomics and metabolomics. The partner project, located in Magdeburg, develops solutions for the analysis of metaproteome data. The BiGi service center concentrates on the computational analysis of isolated microbes as well as of microbial communities. It is equipped with large-scale computing and storage resources, and the institutes in Bielefeld and Gießen were involved in establishing a dedicated de.NBI Cloud infrastructure.
Since the establishment of the de.NBI Cloud 17,18 , all partners work on different cloud-based applications such as tailor-made images for specific bioinformatics analysis (Metagenomics, Nanopore sequencing, ASA³P 19 ) that also play a key role in many of the workshops provided.
Training and outreach activities. Researchers, who are interested in the analysis of genomic, metagenomic or postgenomic data or who want to get familiar with the use of the de.NBI Cloud 17,18 are welcome to join one of the workshops offered by BiGi.
1. Nanopore best practice workshop. To keep up with recent developments in sequencing techniques, a Nanopore best practice workshop has been launched in Bielefeld in 2017 with the support of the service center GCBN that will be held annually due to high demand of the scientific community. The aim of this workshop is to familiarize the participants with the Nanopore sequencing technology, its applications and the 'Best Practice' bioinformatics workflow. The Nanopore technology has greatly facilitated the assembly of prokaryotic and eukaryotic genomes. Therefore, the workshop concentrates on the establishment of finalized genome sequences.
2. Training on genomics tools. Since the start of the de.NBI network, Gießen offers an annual genome analysis workshop. Topic of this three-day-workshop is microbial sequence data analysis including quality control, assembly, genome annotation and comparative genomics with a focus on the usage of the BiGi software tools ASA 3 P 19 and EDGAR 12-14 as well as the BiGi Galaxy Server.
3. Training on metagenomics tools. A further annual course organized by Bielefeld and Gießen together is the introduction into targeted and untargeted metagenome analysis. This course teaches best practices for targeted (16S rDNA operon gene amplicons) as well as untargeted (whole-genome shotgun) metagenome analysis based on high throughput next-generation sequencing data.
4. Training on postgenomics tools. For post-genomic analyses, there is a workshop on the analysis, visualization and integration of multi-level -omics data, which introduces the web-based platform Omics-Fusion 20 . Aim of this workshop is to give answers to typical questions in omics data analyses, e.g. regarding data normalization strategies and handling of missing values, the detection of groups of transcripts, proteins and/or metabolites with similar patterns of expression/abundance using cluster analyses, the visualization of multiomics data in the context of metabolic pathways, or the identification of differentially regulated transcripts, proteins, and metabolites.

5.
Training on metaproteomics tools. A workshop on the MetaProteomeAnalyzer 21 , also including a wet-lab part in some courses, is provided at least once a year in Magdeburg. The learning goal is to apply a complete workflow starting from design of experiments via sample preparation to measurement and bioinformatics analysis of high-resolution mass spectrometry (MS) data.
6. Introduction to the de.NBI Cloud. There are basic courses on the use of the de.NBI Cloud 17,18 , which are offered in Bielefeld and Gießen or also as dedicated workshops on scientific conferences. In these courses, participants learn to setup a research project on the de.NBI Cloud, to work with virtual instances, to efficiently utilize cloud computing resources, about networking and security issues and means of deploying bioinformatics tools in the cloud. The LIFS partner project mainly works on bioinformatics for targeted and comparative lipidomics with applications in cardiovascular and platelet-related disease characterization including lipid identification and discovery 28 . A further focus is placed on lipidomics data standardization and the establishing of a lipid-centric repository for high-resolution mass spectrometry data. LIFS develops and maintains the tools LipidXplorer 29 for shotgun and LipidCreator for LC-MS lipidomics, as well as LipidCompass as a quantitative reference database for tissue and organism-specific lipidomes, and LUX Score 30 for comparative visualization of lipidomes based on their shared lipid structural space.
Training and outreach activities. BioInfra.Prot offers courses mainly focused on mass-spectrometric data analysis and integration in proteomics and lipidomics. Therefore, the courses are mainly aimed either at researchers coming from life or analytical science and working with mass spectrometry data, who want to learn the basic tools and workflows of data preprocessing, statistical analysis and integration, or at bioinformaticians, who want to extend their knowledge on available tools and workflows in these areas.
1. Training on proteomics tools. BioInfra.Prot offers regular courses on bioinformatics for proteomics at the annual German Society for Mass Spectrometry (DGMS) meeting and at Ruhr-University Bochum. These courses are generally open to anyone, providing material for beginners and more advanced participants alike. The online announcement of each course includes information on recommended prerequisites.
2. Training on biostatistics / statistical analysis. Bio-Infra.Prot also offers trainings for differential analysis of quantitative proteomics data and for general introductions to biostatistics and statistical analysis with R. These courses also target beginners and more advanced participants. New courses are regularly announced via the de.NBI portal and on the BioInfra.Prot website. Research focus. CIBI delivers cutting-edge software solutions covering a large domain in the field of computational biology, biomedicine and bioimaging as well as their integration into the workflow system KNIME Analytics platform 33 to enable data-driven innovations and achievements in these fields. Concrete, the service center develops state-of-the-art and strongly competitive tools in the fields of proteomics and metabolomics, genomics, image processing, data mining and workflow integration, with the goal of reducing time and cost expenses for various bioinformatics and biomedical data analysis tasks.

Training on lipidomics tools. LIFS offers an annual
Training and outreach activities. Within CIBI, a broad spectrum of training activities is offered to cover on the one hand the large domain the service center is actively contributing to and secondly to offer distinct training events for different target groups, such as data scientists dealing with big omics-data or application developers and bioinformaticians developing new tools using our resourceful libraries.
1. Training on mass spectrometry data. CIBI offers various courses to introduce mass spectrometry data analysis with a focus on proteomics and metabolomics. In these workshops, concepts such as non-targeted label-free analysis are taught and users learn how to implement complex analysis workflows based on OpenMS tools with subsequent visualisation of the results based on real-life data. In addition to quantification analysis, metabolite identification with MetFrag 34 and MetFamily are taught on several events. The target audience are mainly beginners and intermediate users of proteomics and metabolomics that want to learn how to efficiently process their mass spectrometry data.
2. Training on sequencing data. Another major field in which CIBI offers various training activities relates to sequencing data analysis. Training courses are focused on two user groups. The first are bioinformaticians developing new tools. In dedicated hands-on sessions on the annual user meeting, bioinformaticians are taught how to use the SeqAn software library to write new competitive tools using state-of-the-art algorithms and data structures. The offered trainings mostly focus on beginners but require an intermediate degree of programming knowledge. Knowledge in C++ is helpful but not required. For advanced SeqAn users, we host an annual developer meeting where we tackle specific problems of the participants and add new components to the library. The second target group are data scientists. In these courses, data scientists explore typical analysis pipelines using SeqAn and external tools integrated into the KNIME analytics platform.
3. Training on bioimage data. Annual hands-on courses introduce participants to develop within and for the Fiji 35 , ImageJ2 36,37 , and KNIME 38 ecosystems. In addition, CIBI offers a deep learning course especially for image-based problems.

Training on workflows and tool integration.
CIBI organizes life science workshops on the KNIME Spring Summit and teaches the integration of our tools in KNIME to produce strong and efficient multi-omics pipelines on various meetings (either on international conferences or on joint user meetings). Hence, many of the CIBI events are accompanied with experts in KNIME and users are always welcome to bring their own data so CIBI experts can develop an efficient solution for their problem at hand.

5.
Training on conferences. CIBI offers one-day workshops on international conferences, where the CIBI members can get in touch with life scientists around the world and discuss the topical research subjects and how to tackle them with our software portfolios.
6. Online training. CIBI made most of the training materials available online as self-paced training. These include written tutorials and videos with many practical tips and examples. The materials are intended for selfregulated learning to study and explore the capabilities of our software and tools in a comfortable way. The target audience of these materials ranges from novices to advanced users. The center maintains and develops the largest Galaxy instance in Europe with more than 12000 users, responsible for more than 10 million submitted jobs. The Galaxy server is freely open to all researchers via useGalaxy.eu. RBC maintains an integrated, easily accessible RNA analysis workbench 43 , based on Galaxy, which can be downloaded and installed locally, or deployed on HPC-like environments or clouds. It includes more than 50 RNA tools, multiple workflows, interactive tours and training data.
Training and outreach activities. With its wider focus on the analysis of high-throughput data in Galaxy, an important goal of the RBC is to support researchers by educating them in big data analysis, programming, data management, Galaxy server administration and more. Training and outreach activities. The BioData service center concentrates on data products as well as on research and services. It is well known for its reference databases for taxonomy, phylogeny, biotechnology, biochemistry, pharmacy, medicine, quality control, diagnostics, environmental and biodiversity research. BioData provides analysis services for enzyme structures and functions, classification of next-generation sequencing data. In addition, BioData is highly active in data mobilization and publishing as well as research data management following the FAIR 55 (Findable, Accessible, Interoperable, Reusable) data principles. Therefore, the targeted audience are researchers in the life-or medical sciences, who like to use data and data products as references in their research or applications and/or manage and publish their data in the spirit of science 2.0 aka open-science.
1. Training on FAIR data. The BioData training courses include the topic 'FAIR data', especially within the de.NBI summer school 2018. Here, the courses raise awareness for the importance of proper research data management in general, but also provide a practical toolbox for the acquisition, curation, documentation, archiving and publication of research data following the FAIR data principles. Furthermore, practical training on accessing the data for re-use and integration are part of their training courses.
2. Training on databases. The BioData service center bundles large databases and provides the corresponding training courses for these databases. For instance, in the training course on the rDNA reference database SILVA, in addition, all necessary steps researchers need to take when performing amplicon-based investigations using the rDNA as a marker gene are explained. The course starts with experimental design, including an overview of relevant sequencing technologies, and the selection, design, and evaluation of primers for the amplification of rDNA. It provides examples and best practice solutions for data pre-processing and quality assurance up to the contextualized submission of the NGS data to public repositories. It provides a framework for statistical analysis of the data including a short Linux and R crash course.
In addition, this course also combines an introduction to the BacDive database (Bacterial Diversity Metadatabase) as a tool for further data integration to better understand the biology of sequences/organisms analyzed.
de.NBI Systems Biology Service center (de.NBI-SysBio) de.NBI Systems Biology service center (de.NBI-SysBio) is the service center supporting particularly the systems biology community by providing data and model management tools, manually curated reaction kinetics data and world-wide recognized tools for mathematical modeling of biological networks. The data management part of de.NBI-SysBio is provided by the HITS (Heidelberg Institute for Theoretical Studies). The partner project de.NBI-ModSim (Heidelberg University and the Max Planck Institute for Dynamics of Complex Technical Systems Magdeburg) deliver support in model construction.

Research focus.
The data management part of de.NBI-SysBio has the mission to serve customers with standardized highquality scientific data and data services. The data management system "SEEK for Science" has been developed by the transnational FAIRDOM project 5 as a sharing space for projects fitted to the needs of systems and synthetic biology as well as systems medicine. The database SABIO-RK provides structured, standardized and annotated kinetic data with a focus on supporting the computational modelling community to create models of biochemical reaction networks as well as on allowing experimentalists to gain further knowledge about enzymatic activities and reaction properties. The data coming mainly from literature are expert-curated and can be easily exported for direct use, e.g. in modeling tools like COPASI 54 to run simulations.
The modeling part of de.NBI-SysBio is provided by Heidelberg University and the Max Planck Institute for Dynamics of Complex Technical Systems Magdeburg and their tools COPASI 56 and CellNetAnalyzer 57 . CellNetAnalyzer (CNA) is a MATLAB toolbox serving a graphical user interface and various computational methods and algorithms for exploring structural and functional properties of metabolic, signaling, and regulatory networks. COPASI is a widely used software tool for creating, simulating and analysing models of biochemical reaction networks.
Training and outreach activities. The de.NBI-SysBio training courses, workshops and tutorials are suitable for beginners as well as for advanced users. For instance, the three-days training course 'Tools for systems biology modeling and data exchange: COPASI, CellNetAnalyzer, SABIO-RK, SEEK' with extensive hands-on sessions annually takes place alternately in Heidelberg or Magdeburg. Additionally, online tutorials are offered to introduce the main tools (e.g. FAIRDOMHub 5 , SABIO-RK 58 , COPASI 56 , CellNetAnalyzer 57 ).
As regular international outreach activities de.NBI-SysBio is involved in the organization of satellites of the annual ICSB conference (International Conference on Systems Biology) with the 'COMBINE Tutorial -Modelling and Simulation Tools in Systems Biology" and "Advanced Modeling with COPASI'.

Training on data management and FAIR data.
de.NBI-SysBio was involved in organization of the de.NBI summer school on FAIR data and data management in 2018. The data management training courses are open to everyone interested in producing FAIR data in project driven scientific research. de.NBI-SysBio offers special training courses for experimentalists or for modelers but also more general workshops and tutorials with breakout sessions to delve into the topic from a modeler's, experimentalist's or developer's point of view (e.g. annual Systems Biology Developers Foundry).
Furthermore, de.NBI-SysBio offers on demand visits to support customers in installing their projects or local SEEK instance.
2. Training on modeling. The trainings for the modeling tools CellNetAnalyzer and COPASI offered by de.NBI-ModSim aim at modelers or those who want to become one. Attendees learn basic model construction and analysis techniques for kinetic modeling of biochemical systems (illustrated and exercised with COPASI) and principles of stoichiometric and constraint-based modeling of metabolic networks (coupled with hands-on exercises using CellNetAnalyzer).

de.NBI Cloud-based training
In order to perform analysis on datasets available in the life sciences, an appropriate compute infrastructure is necessary. The de.NBI Cloud was created to fill this gap and offers storage and compute resources for researchers in life science in Germany. Through a cloud federation setup, the six cloud sites including Bielefeld, Gießen, Freiburg, Heidelberg, Berlin and Tübingen are integrated into a single cloud platform and offer more than 33000 cores and 38 PB of storage capacity in total. A single sign-on (SSO) mechanism enables the user to access any de.NBI Cloud service via their home institution account using the ELIXIR authentication and authorization infrastructure system (ELIXIR AAI) and in particular Perun, the identity and access management system of ELIXIR 17 .

Technical aspects
Currently, the de.NBI Cloud offers the project types SimpleVM and OpenStack 17 . Both project types are using virtual machines (VM), where a VM is an emulation of an operating system and represents one of the main building blocks of the cloud. Open-Stack is an Infrastructure as a Service (IAAS) system that allows a developer to configure any computational resources like networking, storage and VM settings for running largescale analyses or offering of a web service in the cloud for other researchers. SimpleVM is a beginner friendly project type that streamlines the handling of VMs and does not demand any particular knowledge in Cloud Computing. While OpenStack is suitable for training in the context of Big Data like running workflows for metagenome analysis, the SimpleVM project type can be used for command line, Linux or any other trainings, where running tools on one virtual machine is sufficient.

Application process and getting started
The procedure for applying, using and accessing computational resources for a de.NBI Cloud training course is the following. First, a principal investigator (PI) of a German university or research institution must apply for a project through the de.NBI Cloud portal. As soon as the access committee approves the application, the PI can add the actual trainer and set him as an administrator of the project. The trainer first starts a VM by choosing an appropriate Linux system, e.g. Ubuntu or CentOS, and then installs any software needed by the course participants. Once the trainer has installed the required software, the VM can be snapshotted and made accessible to all course participants. At the beginning of the training course, the PI adds further course participants, who can use exactly the same tools based on the initial VM created by the trainer. Finally, the trainer and the participants can directly start with the actual course.

Training activities
Since the beginning of 2018, the de.NBI Cloud is in production and multiple workshops, tutorials and even a user meeting have been organized. The given workshops range from research related topics like Nanopore workshops to more cloud computing related topics.
1. Training on sequencing data. In Nanopore best practice workshops participants are taught about Nanopore sequencing technology, its applications and the "Best Practice" bioinformatics workflow (see subsections on BiGi and GCBN). Nanopore tools such as long read assemblers demand many cores and especially a lot of RAM that can be made available by the de.NBI Cloud.
2. Training on metagenomics tools. Other training events like metagenome workshops are demonstrating how to store and analyze large datasets using state of the art bioinformatic tools in the cloud (see subsection on BiGi). These workshops highlight one of the main use cases of the cloud, which is the handling of Big Data. Tool-and topic-specific training was also extended in the last few years (Table 1) Regarding performance, de.NBI and ELIXIR-DE guarantee high quality standards (~90% recommendation rate, ~86% very good or excellent votes based on about 2000 feedback responses -July 2020) above ELIXIR average (~89% recommendation rate, ~70% very good or excellent votes based on about 8500 feedback responses -July 2020) ( Figure 2). However, a general comparison of the training programs provided by each ELIXIR node is very difficult. Every country has a specific teaching focus and different amounts of training courses. In general, ELIXIR UK and ELIXIR Switzerland offer a broad portfolio of training courses similar to de.NBI / ELIXIR-DE 60,61 , whereas smaller ELIXIR nodes, like Estonia, are only able to offer a few courses including the main research aspects in their countries.  In addition to face-to-face training, online training was introduced on the de.NBI website in 2016. Further, online hackathons for different software packages were established by the service center RBC in 2016. In 2019, the repertoire of online material was increased to 45 items and guidelines for online training were developed by SIG 3. In 2020, further online material will be provided by the different service centers to reach about 60 items -and thus a larger audience that cannot be covered by the training courses alone.

Summer schools and user meetings. Cloud
To sum up the effort of the de.NBI / ELIXIR-DE training platform: Since 2015, de.NBI service centers, the de.NBI Administration Office and SIG 3 worked very successfully together to establish, coordinate and expand bioinformatics training courses across Germany and in ELIXIR. In total, about 6,000 participants (Status July 2020) were trained in different bioinformatics disciplines to tackle the challenges of the five Vs of the big data problem.
Nevertheless, training still has a high priority for de.NBI / ELIXIR-DE and significant future efforts will be put into the creation of more e-learning material, the recruitment of new external de.NBI training partners and the qualification of trainers to keep pushing training activities forward and to continue participating in the ELIXIR training platform.

Data availability
No data are associated with this article.

Open Peer Review
Introduction to the de.NBI / ELIXIR-DE training platform: "Trainer meeting" --> "Trainer meetings" Figure 1: What is the meaning of the circle's size? In my opinion, this entire section should be moved elsewhere, as it doesn't seem to fit in the "Cloud-based training" section. In particular, in the current (revised) version of the paper, its subject has already been partially addressed in the sections "Introduction to ELIXIR-DE" and "Introduction to the de.NBI / ELIXIR-DE training platform", so a possibility could be to integrate its content within these sections. Future plans and conclusion: ... "above ELIXIR average" should be replaced by "equal or superior to ELIXIR average" given that, regarding the recommendation rate, the percentage provided by the ELIXIR Training Metrics Database (up to the end of July 2020) is 89,8%. Therefore, the difference between the ELIXIR-DE value and the ELIXIR value seems not significant.
In Future plans and conclusion: Further --> Furthermore   Etc... Narrative could be used to describe key features of the platform (as listed above) and of courses. e-learning model(s) also could be better described (synchronous?asynchronous? Blended? etc). A description of service centres and their research foci could be provided very succinctly in Figure 1 or in a separate table (listing, for each centre, the research focus and member institutes) and then leave further details to the literature reference.
Thank you for the comment. In general, our courses are on an annual basis. Therefore, we provide in 80 courses almost 80 different types of courses. Therefore, such a table would be too large to present in the paper. The service centers have unique topics in de.NBI / ELIXIR-DE. Therefore, there is almost no topic overlap in the training courses between the different service centers. That is the reason, why we choose this strategy to present the training program. To describe the training platform in detail, we included an additional part introducing the training platform.
3) Figure 1  Thank you for the comment. The circle size is now explained in the figure legend. The service centers represent the training topics. As explained before, each service center has its topic and there is almost no overlap with another center.

Specific comments/questions: What do you mean by "Big Data Problem"? Could you add a reference and explain what characterises it as a "problem"? Was added
In the abstract, we suggest that you replace "European ELIXIR network" with "European ELIXIR infrastructure".

What do you mean by "standardisation of training course monitoring"?
We developed a standardized form that must be filled out after every course and user survey based the ELIXIR survey questions.

A sentence on what is TeSS should be written before mentioning it in the Introduction.
A sentence on TeSS was added.
p. 4: Re: "Qualification of more trainers." How do you qualify trainers? What is a "qualified trainer"?
We plan to organize more Train-the-Trainer and Software Carpentry Instructor course. In addition, a trainer community meeting is planned.
As SIG3 is for "Training & Education", do you also have programs for undergraduates and high school students?
No, we don't have a special program for undergraduates and high school students. About 15% of all participants are undergraduates and we support a summer school for high school students.
In the abstract, the overall number of participants is updated to October 2019 ("since 2015, more than 250 training courses were carried out with more than 5,200 participants"), whereas in the paper it is updated to July 2019 ([...] more than 4800 participants (Status July 2019)). This data should be made consistent.
Was updated to July 2020. Reference was added delivering ~200 courses. The discussion of lessons learned is vary valuable to course providers in other infrastructures (such as me). I acknowledge that such critical reviews might not find universal support across the many people involved in the training activities in de.NBI.
The structure of the overview can be improved by shortening the sections devoted to the individual centers and instead discussing items that are taught across all centers, such as basic statistics courses.
The contribution from the individual centers are too long and frequently include vague or obvious statements (existence of an HPC facility, the online announcement of course and requirements). A stronger editing mandate would be advised.
Software developed in a particular center for which training is provided should only be mentioned in the section that discusses the training. There is not usually need to repeat them multiple times in the text.
Figures: Figure 1. The composition of the individual centers by cities is difficult even for someone acquainted with the geography of Germany. There is space to include the names of the centers, which could then receive less coverage in the text. ○ Figure 2. Has no content, would recommend to delete. The geographic spread of de.NBI nodes using AAI provides no insight beyond the names. Thank you for the comment. Sometimes, the software needs to be mentioned to underline the research focus of the center. Repeats were reduced.
Figures: Figure 1. The composition of the individual centers by cities is difficult even for someone acquainted with the geography of Germany. There is space to include the names of the centers, which could then receive less coverage in the text.
The names of the service centers are mentioned on the left side of the figure. The color underlines the belonging to a service center    Figure 2).
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com