Lab in a box: A build-your- own-open-lab software toolkit

Over the past two years, our team has been working on an interoperable software toolstack that is open source, self-hosted, and covers basic relevant needs of a computational neuroscience lab.Notably, a number of software solutions came into existance or were deployed or further developed thanks to interactions of different software communities during RDM workshops or the distribits conference for distributed data management technologies (distribits.live).Our objective is to design an approach that allows storing data of arbitrary size, flexible semantic meta data, and the relations between these data; and to provide ways to query those relations and access the underlying data, as well as exposing selected data for websites, knowledge bases, or data catalogs.The system components are either fully compatible or integrated with the DataLad (Halchenko et al., 2021) ecosystem for data management.At the core of the stack, we have developed the following software components: Forgejo-aneksajo1: A lightweight collaboration hub for code and data thanks to native git-annex support. It has a comparable feature set to GitHub (CI/CD, private versus public repositories, container registry, …), but allows further customization (including, for example, nii-vue integrations for DICOM and Nifti previews, self-hosted runners for CI/CD actions, self-hosted container registries, …). Shacl-vue2 and dumpthings3: The front- and backend for a metadata annotation stack, tuned for flexibly and user-friendly collecting, curating, storing, and quering metadata. datalad-concepts4: A set of pragmatic meta data schemas that enable alignment with arbitrary vocabularies. The schemas feed directly into the metadata editor, and can also be easily extended. In addition, a set of existing open source softwares complements the stack:Hugo-based website scaffolds can be bootstrapped from or extended with meta data from the meta data stack.Hedgedoc, with authentication via a forgejo-annexajo hub, constitutes a feature-rich collaborative editor.And a CalDAV-based calendar and task service provides custom calendars and shared todo lists.While the components are all standalone and can be deployed individually, they are tightly integrated and interopable, enabling additional usecases when combined.For example, the metadata annotation stack gains permission management and document upload routines when used together with forgejo-aneksajo.In addition, the system is an optimal fit around the DataLad tool for data versioning and publication.Importantly, the deployment for each service is automated using pyinfra5 to make initial setup, configuration and maintenance effortless, even for scientists without experience in IT administration. In conclusion, we have created a lab-in-a-box toolkit with self-hostable, open source software solutions for the digital service needs of an open neuroscience lab.All resources are openly available and can be found at https://hub.psychoinformatics.de/lab-in-a-box/liab-deploymentsThe stack strengthens the digital sovereignty and resilience by replacing commercial services whose available hinges on for-profit companies and that can be vulnerable to political or technical instability.Successful roll-out in different neuroscientific contexts have proven its initial utility, and ensure continuous improvement through co-development and user feedback. https://codeberg.org/forgejo-aneksajo/forgejo-aneksajo ↩︎ https://github.com/psychoinformatics-de/shacl-vue ↩︎ https://github.com/christian-monch/dump-things-server ↩︎ https://concepts.datalad.org/ ↩︎ https://hub.datalad.org/infra/deployments ↩︎