data2health / cloud-tool-architecture Goto Github PK

Starting this thread to start capturing ideas. These will constitute a feature matrix that NCATS can use to assess/evaluate GitHub and GitLab.

• Public (documented) API — ability to programmatically interact with platform
• Project management — (a la Agile) issue tracking and management, visualization of work
• Private repos — ability to create private repos and share with limited teams/repos
• Multi-repo organization — ability to logically group and manage (through UI or workflows) related repos
• Community features — ability to associate repos, users, and teams with an organization (e.g., NCATS)
• Issue templating — ability to specify the content, structure, and format of an issue/ticket
• Pull/merge request templating — ability to specify the content, structure, and format of a merge request description
• Licensing — free or low-cost options for academic and non-profit institutions
• CI/CD integration — support for build/test/deploy pipelines and workflows through built-in or third party services
• Webhook/app integration — ability to connect third party apps/services
• Federated auth (SSO) — federated authentication to access repos (?) and connected services (?)

Anything I'm missing?

https://duos.broadinstitute.org/#/home
https://github.com/DataBiosphere/consent-data-use

Based on the 9/30 call, some items that we want to address (and will need to iterate on) include:

• Governance (of the tool registry, of hosted tools, etc.)
• SOPs (for development, deployment)
• UI design (landing page, organization, common themes/elements)
• Infrastructure implementation (what do we build ourselves vs. adopt/extend — CyVerse is a key focus here)
• UX design (processes and features for "registering" new tools, apps, etc.)

Not sure where these should go: Google Drive, markdown in this repo, issues and comments in this repo, additional specialized repos, etc.?

Find and add examples of creating container images for models/algorithms in at least R and Python (maybe Perl). Part of the instructions should include recommendations for base image, depending on language and use case.