Skip to content

Updating

Application updates

Yeliztli checks GitHub for new releases at startup (configurable via update_check_interval). When an update is available, a subtle indicator appears in the UI.

To update a native install:

cd Yeliztli
git pull
pip install -e .
cd frontend && npm install && npm run build && cd ..
yeliztli-setup install   # re-register and restart services

For Docker, rebuild and recreate the containers:

git pull
docker compose up -d --build

Back up first

Snapshot your data before a significant update — see backup & restore.

Reference-database updates

Reference data updates are managed from Settings → Database Management for the sources registered in the update manager (see reference data). That panel currently tracks ClinVar, dbNSFP, CPIC, GWAS Catalog, dbSNP, MONDO/HPO, ENCODE cCREs, and the published bundles for gnomAD, VEP, LAI, and PGS scores, plus the app-shipped ancestry PCA bundle. There you can configure:

  • per-database auto-update toggles,
  • update check frequency (startup / daily / weekly),
  • an optional bandwidth window for large downloads, given as a 24-hour "HH:MM-HH:MM" range (e.g. "02:00-06:00"). The updates route holds large downloads to that window; a Force update action bypasses it when you need a download now.

Update history is logged and viewable in the Settings panel. A few installed or optional reference sources are outside that update-manager registry: AlphaMissense, GTEx eQTL, ClinGen, and bring-your-own SpliceAI. Yeliztli does not check upstream releases for those sources, does not show them in the per-database auto-update table, and cannot auto-refresh them; refresh them only through the relevant manual rebuild, local ingest, or setup flow.

How updates affect your existing results

Your findings are a point-in-time snapshot: each sample's results are computed against the reference data installed when that sample was annotated. Updating reference data afterwards does not change results you already have.

  • Updates don't re-annotate automatically. Downloading newer ClinVar / gnomAD / etc. data leaves your existing samples' findings untouched until you re-annotate them.
  • Reference-data staleness prompts are broad and neutral. After a successful annotation, Yeliztli records the reference-database versions used for that sample. When any installed reference database later moves beyond that snapshot, Settings shows a single per-sample re-annotation indicator: "Reference data is newer than this analysis." Most database updates will not change a given sample's findings; the prompt means the analysis snapshot is behind, not that a clinical interpretation changed.
  • ClinVar reclassification prompts remain specific. Yeliztli still separately flags ClinVar significance changes that affect one of a sample's variants (or a variant you are watching). Those prompts identify potential reclassifications; the broader reference-data prompt only says the sample should be re-annotated to refresh against newer sources such as gnomAD allele frequencies, dbNSFP predictions, CPIC prescribing guidance, the GWAS catalog, ENCODE cCREs, or the VEP/LAI/ancestry/PGS bundles. Sources outside the update-manager registry can participate only after you manually rebuild or ingest them and their installed version stamp changes; they are not proactively checked for upstream releases.
  • Re-annotate to refresh. To bring a sample's findings up to date after any reference-data update, re-annotate it: accept the re-annotation prompt when one appears, or open that sample's dashboard and click Run Annotation in the Annotation Pipeline panel. If the panel still shows a completed, failed, or cancelled status, dismiss that status first. The status bar's reference-database versions (see reading your results) show which snapshot a result currently reflects.

After updating databases, re-annotate to refresh findings

A reference-database update only affects new analyses. Existing findings keep showing the older data until you re-annotate the sample.