SeqAuto run status page - user-facing pipeline visibility, grown from the dashboard

🤖 Written by Claude

## Problem

SA Path VG3 lab users have no visibility between "run appeared in the sequencing runs list" and "my sample finally shows variants". The annotation pipelines page is admin-only, and VCF ingest is 100% API (the upload UI got 14 hits in 15 days of production logs), so import and annotation progress is a black box. Production logs (4–18 Aug 2026) show users polling for readiness anyway: the sequencing runs list loads ~47×/day and the variantopedia dashboard gets ~35 visits/day.

The dashboard (`variantopedia/views.py` `dashboard`, `dashboard.html`) is already a proto-version of the needed page: its `SEQAUTO_ENABLED` branch shows "Recently Uploaded Sequencing Data" (run → sample sheet → samples with links) plus the samples-by-kit graph — and has an *implicit* readiness signal (sample names only become links once `SampleFromSequencingSample` exists). It's mostly seqauto-specific code living in variantopedia.

## Proposal

**1. Move the dashboard's seqauto branch into a seqauto "run status" page** and grow it there:

* Newest-first list of recent sequencing runs (paged/datatable, not the hardcoded 5)
* Run-centric rows matching how the lab thinks ("is Haem_26_075 ready?"), with explicit pipeline stages replacing the implicit link-vs-no-link signal:
* **Arrived** — run/sample sheet ingested via API
* **VCF received** — combined/joint-called VCF uploaded
* **Importing** — UploadPipeline state
* **Annotating** — AnnotationRun progress (variants annotated vs total, per build), with the rough ETA ideas from #998 (historical per-version throughput / VEP output monitoring)
* **Ready** — link straight to sample / auto-created analysis
* Run-level rollup badge (e.g. "annotating 3/8") plus per-sample state within the row
* Auto-refresh while runs are in flight (users leave pages open all day)
* Per-run drill-down for fine granularity: that import's upload pipeline steps and the annotation run(s) its variants landed in — read-only, scoped to the run clicked, so users get detail along an axis they can navigate (their run, not global AnnotationRuns). Admins get a link through to the annotation pipelines page.

**2. Redirect `variantopedia/dashboard` to it when `SEQAUTO_ENABLED`**, keeping the generic dashboard (samples-by-month graph, changelog, notices) for non-seqauto deployments (Shariant, variantgrid.com).

**3. Make it the seqauto index page** — so `/seqauto/` and the dashboard converge on one habit instead of two.

## Related

* #998 — annotation pipeline page stays the admin/ops drill-down; the ETA/progress ideas there feed this page's "Annotating" stage. Note the user-visible need lands here, not there.
* #1748 — seqauto run list/detail improvements; this page becomes the landing that links into those.

Usage evidence in `claude/vg3_usage/` (VG3 production nginx analysis).

Start with variantopedia/views.py and dashboard.html, especially the SEQAUTO_ENABLED branch, then review the existing seqauto run/detail work referenced by #1748 and the annotation progress ideas in #998. The work is done when seqauto has a run-status landing page with staged run and sample visibility, drill-down links, refresh behavior, and the dashboard redirects to it for seqauto deployments while retaining the generic dashboard elsewhere.

Written by the indexing model from the issue text.