Recommendations for hosting model code with DOI

[metazool]

We had a query forwarded from the Data Management team about recommendations for hosting model code.

• Cross-institutional partnership, and "external" contributors need repository admin rights
• Reasonable guarantee of long-term preservation of the repository

"Is this something we can do on the UKCEH github account?"

What do we recommend here and what questions do we need to ask? (I've definitely got one about software licensing; if the partners share "IPR" in the model, what does that imply for code reuse?)

There's a more general request from Data Management for a standard pattern for hosting model code repositories, providing a DOI and links to entries in the model catalogue.

Is this a case where a "Github Organisation" specific to model repositories would make sense?
If we took Github hosting out of the framing of how this should work, would it look different?

https://catalogue.ceh.ac.uk/documents/930ded2c-35b4-422d-80c0-1744f59c3763 - Catalogue record for this specific model (may require login)

https://cehacuk.sharepoint.com/sites/Hub-research-datamanagement/SitePages/Section-ModelManagement.aspx (Guidance as it stands for hosting model code in git - almost certainly requires login)

See also: the wider Discussion about curating the organisational Github presence and our team's role in that #44

[metazool]

@EdwinRowe I think this query came from you originally? Is there any detail you'd wish to add?

I'm not sure what the policy is, or the technical constraints are, around adding external contributors as project Owners in our organisation - as a minimum, we might require that they have got two-factor authentication enabled? It's something to check with @DamienUKCEH or other owners of the NERC-CEH Github organisation

The repository can always be moved if there's a more suitable home for it identified - as you've already got the membership set up, you could go ahead and create repo here for your model code, and we can treat this as a case study for handling it in a standard way, rather than waiting for the meta-discussion about curating the organisational Github presence to fully resolve (it could go on for as long as UKCEH does :D)

[EdwinRowe]

@metazool that's very helpful, thanks. I'll set up the repo here and see how we get on. The draft MoA that we're working on with Lancaster and Liverpool is below - comments welcome. It would be brilliant if we could provide a DOI. How would that work with versions, would the DOI refer to a particular commit?

Memorandum of Agreement on the use of the N14CP model of ecosystem biogeochemistry:

1. The N14CP model (the Model) is described in Schedule 1 of this Agreement and includes contributions from authors from UKCEH, Lancaster University and University of Liverpool.
   • Janes-Bassett V, Davies J, Rowe EC and Tipping E (2020) Simulating long-term carbon nitrogen and phosphorus biogeochemical cycling in agricultural environments. Science of the Total Environment 714: 136599.
   • Davies JAC, Tipping E, Rowe EC, Boyle JF, Graf-Pannatier E & Martinsen V (2016) Long-term P weathering and recent N deposition control contemporary plant-soil C, N and P. Global Biogeochemical Cycles 30: 231-249.
   • Tipping E, Rowe EC, Evans CD, Mills RTE, Emmett BA & Chaplow JS. (2012) N14C: a plant-soil nitrogen and carbon cycling model to simulate long-term ecosystem enrichment by atmospheric N deposition. Ecological Modelling. 247: 11-26.

2. The Model was developed under the LTLS project. In accordance with the licence granted under the LTLS agreement, ownership in copyright for the model is shared between UKCEH, Lancaster University and Liverpool University.

3. All members of the LTLS consortium (including the Parties to this Agreement plus the British Geological Survey, the James Hutton Institute, Rothamsted Research, Keele University and University College London) have been granted an irrevocable, non-transferable, royalty-free right to use the Model.

4. This Agreement sets out the future licence conditions of the Model as agreed by the Parties whereby:

i) A definitive body of code for the current version of the Model has been made available in a Github repository at …

ii) The code is published under the MIT licence (https://opensource.org/license/mit).

iii) Licensees are asked to use the following acknowledgement: N14CP Model © UKCEH; © Lancaster University; © University of Liverpool. [year of publication]

iv) We ask that publications using the Model include a link to the code repository.

[metazool]

It would be brilliant if we could provide a DOI. How would that work with versions, would the DOI refer to a particular commit?

How this typically works is your DOI is associated with a "release" - when you decide your changes are significant enough to merit a new DOI.

The process for creating a release is a manual one - links in that page above, or directly here to "managing a releases in a repository"

The RMAVIS repository looks like a good example of an existing NERC-CEH project that is set up to issue a DOI for each release in this way.

I'll refer you to colleagues @longr and @jmarshrossney for any followup on setting up your model repository, as I'm about to head off on a long period of annual leave ... good luck with it!

[metazool]

@dcmo writes the current advice from the NERC Data Centres (including EIDC) is that researchers should put their model code in in GitHub and lodge the code with Zenodo (and let Zenodo worry about the issue of long-term currency). I feel UKCEH should have a long-term GitHub-based solution for staff especially for models/code that are continually maintained and updated by teams rather than individuals.

If we were planning to set up an Organisation dedicated only to models, who else would one need to consult? This probably belongs in a Discussion somewhere :D