Hi Cliburn,

As you know, MDSC is in the process of writing a manuscript evaluating existing FAIR (Findability, Accessibility, Interoperability, and Reproducibility) principles for models, specifically in terms of IID models.

We would like to collect feedback from the RPs on how (or how not) Findability, Accessibility, Interoperability, and Reproducibility show up in their domains for mechanistic models, ABMs, and ML/AI models. Would you be able to help circulate this to the RPs and request a response by Friday, March 20?

A couple paragraphs from each RP would be great. I’ve included some example questions and resources below to stimulate thoughts.

How easy or difficult is it to find, access, and run models from other researchers?
Are there certain repositories used to store and share models and datasets?
How/where do you typically find relevant models and datasets?

Are there certain formats used to increase interoperability and reusability such that models can be reused and integrated in other models/systems?

Machine-readable formats including SBML, CellML, etc
Modular code or resources
Containerization

Is annotation typically included for understandability and interoperability?

Using ontologies (EDAM, SBO, etc)

Are there differences in the FAIRness of sharing mechanistic models versus agent-based models versus ML/AI models?

Relevant resources:

read more about the FAIR principles for research software
This paper discusses further FAIR guidelines for mechanistic models.

Best,

Aubrie

Aubrie Weyhmiller
Research Scientist
Renaissance Computing Institute (RENCI)
Pronouns: she/her