GeoEpi Hub

What belongs in the operational lab-level project portfolio.

GeoEpi maintains a lab-level GeoEpi Hub in the GeoEpi GitHub organization. The Hub provides shared project context, data-stewardship records, canonical subproject registries, and automatically generated portfolio status. It is not an analysis repository or an HPC run registry.

Information stored directly in the Hub must be appropriate for everyone who has access to that repository. GitHub repository permissions generally do not provide project-folder-specific access control.

What a project entry contains

An individual folder in the Hub is a project entry. It should make the project understandable to collaborators who do not have SCINet or HPC access. A project entry normally records:

  • project identity and objective;
  • project leads and collaborators;
  • milestones and major non-sensitive decisions;
  • major deliverables and product links;
  • links to the project Data Management Plan and controlled-data inventory;
  • the canonical subproject registry; and
  • references to agreements or restricted resources where appropriate.

The Hub describes the project. The subproject repository describes the science. The computational environment performs the work.

Minimal project-entry structure

Keep the project entry deliberately small:

projects/
  nws-risk/
    README.md
    decisions.md
    data/
      data-management-plan.md
      controlled-data-inventory.csv
    subprojects.yml
    products.yml

Complex projects may add scope.md, people.md, milestones.md, or deliverables.md, but these are not required as separate files when the project README already communicates the information clearly. Use the project overview template when creating or adapting a project entry. Not every optional file is needed on day one.

Access boundary

The Hub should not contain controlled data themselves, credentials or tokens, sensitive personnel information, confidential partner communications, restricted agreements when repository access is too broad, or sensitive operational information inappropriate for all Hub users.

Where restricted material exists, record its existence and authoritative approved location when appropriate. The Hub records where restricted resources belong; it does not need to contain the restricted resources themselves. Do not invent institutional security classifications; follow the applicable agreement, security guidance, and institutional direction.

Human-maintained and generated content

The working division is:

projects/                           human-maintained project records
projects/*/subprojects.yml          human-maintained canonical pointers
generated/portfolio-summary.md      automation-owned executive view
generated/attention-needed.md       automation-owned exception view
generated/subproject-status.md      automation-owned detailed view
generated/subproject-status.csv     automation-owned machine-readable view

Files beneath generated/ are built by the Hub synchronization workflow from registered repositories’ .geoepi.yml metadata. They should not normally be edited manually.

Synchronize project state, not project content.

Run parameters, package versions, random seeds, SLURM job IDs, input manifests, logs, output manifests, and detailed analytical decisions stay in the subproject repository or computational storage. The Hub points to canonical repositories and project context; it does not duplicate every run.

See Add a Project or Subproject for the operational registration procedure.