docs.trr379.de/content/goals/rdm-infrasturcture/_index.md

166 lines
6.1 KiB
Markdown

---
title: RDM Infrastructure
weight: 10
mermaidZoom: true
---
**[This is a draft under discussion]**
The planned RDM infrastructure is a federation of interoperable site
infrastructures. The key design principle is that *no primary data are
aggregated to a central infrastructure*.
We aim to establish an infrastructure that is suitable for use within the
TRR379, but not limited to this scope. Once deployed, the associated services
are usable beyond the scope of TRR379.
The following schema sketches the planned infrastructure. Components
that hold (primary) data are depicted in yellow. Components that hold
(mostly or exclusively) metadata are shown in blue. The direction of
information flow is indicated by arrows, exchange of data by solid arrows,
and metadata-only exchange by dotted arrows. Infrastructures that are
only accessible to authorized agents are labeled with a "lock" symbol.
Further details on individual components are provided below.
{{< mermaid >}}
graph TB;
subgraph Central services
C1[("Collaboration portal<br>hub.trr379.de<br> 🔐")]:::meta
C2{{"Data search<br>query.trr379.de<br> 🔐"}}:::meta
C3(Main website<br>www.trr379.de):::meta
C4("Data catalog<br>data.trr379.de"):::meta
end
subgraph "Aachen 🔐"
A1[(hub)]:::data
A1a[(lab1)]:::data
A1b[(lab2)]:::data
A2{{query}}:::meta
end
subgraph "Frankfurt 🔐"
F1[(hub)]:::data
F2{{query}}:::meta
end
subgraph "Heidelberg 🔐"
H1[(hub)]:::data
H1a[(ZI-hub)]:::data
H1b[(lab1)]:::data
H2{{query}}:::meta
end
A1 -.-> C1
A1a -.-> A1
A1b <---> A1
A1 -.-> A2
A2 <-.-> C2
F1 -.-> C1
F2 <-.-> C2
F1 -.-> F2
C3 <-.-> C1
C3 -.-> C2
H1 -.-> C1
H1a <-.-> C1
H1a <-.-> H1
H1b <---> H1
H1 -.-> H2
H2 <-.-> C2
C1 -.-> C2
C1 -.-> C4
C3 -.-> C4
H1b <---> A1a
%% node links to actual services
click C1 href "https://hub.trr379.de"
click C3 href "https://www.trr379.de"
%% classes to distinguish data and metadata nodes
classDef data fill:#ffa200,color:#000
classDef meta fill:#5C99C8,color:#000
%% invisible link purely for manipulating the grouping
C2 ~~~ A1b
C4 ~~~ A1b
C2 ~~~ F1
C2 ~~~ H1b
{{< /mermaid >}}
## Central services
All central services a metadata-focused. No primary data acquired at
participating sites are aggregated into central databases/storage.
### Collaboration portal (hub.trr379.de)
This is the main hub for collecting actionable links to all TRR379 resources
and information. The software solution for this hub is
[Forgejo-ankesajo](https://codeberg.org/matrss/forgejo-aneksajo). It is a free
and open-source software package, and the direct service counterpart of
[DataLad](https://www.datalad.org), the main tool proposed for implementing
reproducible research workflows in TRR379 labs.
The hub will store DataLad datasets, referencing all TRR379 data resources
without hosting any actual data. Instead the DataLad dataset point to the
individual institutional data stores, or to community data repositories when
and where data have been published.
The hub is also a place to deposit (shared) computational environments, and
implementations of (shared) data processing pipelines, software publications,
and source code repositories under a uniform TRR379 umbrella.
A test site is deployed at https://hub.trr379.de and is being evaluated.
### Main website
See [this page](/websites/www/) for a description of the main website.
Importantly, the website renders essential metadata for the TRR379
(contributors, roles, publications, projects, research topics, etc.) It
provides a unique
[URI](https://en.wikipedia.org/wiki/Uniform_Resource_Identifier) for any such
entity, to be used as identifiers in all TRR379 (meta)data resources.
The website is programmatically generated from a [repository hosted on the
TRR379 hub](https://hub.trr379.de/q04/www.trr379.de), to enable contributions
by all TRR379 members.
### Data catalog
The TRR379 data catalog is a website dedicated to providing a uniform
(read-only) view on the TRR's data resources. It is rendered programmatically by
[DataLad Catalog](https://docs.datalad.org/projects/catalog) from metadata on
TRR379 data resources hosts in the TRR379 hub.
This site is indexed by specialized search engines like [Google's dataset
search](https://en.wikipedia.org/wiki/Google_Dataset_Search) and a key enabler
for general findability of TRR379 resources.
### Data search
This is a federated data discovery service that is tailored to the cohort dataset
acquired by TRR379 as a whole. It will enable the discovery of individual data records
matching a given set of criteria, regardless of the contributing TRR379 site.
The service is federated. Each sites runs their own instance, and has the sole
authority on deciding what metadata are shared with other TRR379 sites. Only
these metadata property will be accessible by TRR379 at large, while more
detailed metadata records can be use for in-house queries.
The proposed solution for the query service is a version of
[NeuroBagel](https://query.neurobagel.org) adapted to the data nature and needs
of TRR379.
## Site infrastructure
Sites are free to implement any RDM solutions, as long as that infrastructure
provides
- (programmatically) queriable metadata of a previously agreed upon nature
- (programmatically) accessible data to any authorized members of TRR379
with the aim to enable reproducible research from primary data to published
results within and across TRR379.
[Q02](https://www.trr379.de/projects/q02) supports sites with software solution
that facilitate interoperability within TRR379. This includes the local
deployment of the software systems used to run the central services.
We aim at individual sites running their own data hubs (using the same software
solution) as the central https://hub.trr379.de. In contrast to the central hub,
these institutional sites can directly use the storage features of
[Forgejo-ankesajo](https://codeberg.org/matrss/forgejo-aneksajo), and host
arbitrary amounts of data. TRR379 can communicate data availability using a
federation protocol.