Skip to content

Inject Web-BLAST-parity columns into tabular outfmt

Motivation

Running a search with a tabular output format (-outfmt 6 / -outfmt 7) left the result page's Description, Scientific name, and Query/HSP Cover columns blank. The dashboard's result analytics read those values from named tabular columns — stitle (description), sscinames / staxids (scientific name), and qcovs (query cover) — but the default tabular layout (std, or a bare 6 / 7) carries none of them. An -outfmt 5 (XML) run was rich because XML embeds that information; a tabular run was not.

User-facing change

  • When a submit uses a tabular outfmt, the dashboard now appends the missing result-UI parity columns — staxids sscinames stitle qcovs — at submit time. A tabular run's result page is now as complete as an XML run's.
  • The caller's own columns are preserved and never duplicated (idempotent). A bare 7 becomes 7 std staxids sscinames stitle qcovs; outfmt 5 (XML) and an already-complete layout are untouched.
  • Applies to every tabular submit path — New Search (V1), the Service Bus Playground send, and an external queue producer — because they all build the same ExternalBlastV1Request, where the injection lives.
  • Side effect: a downloaded tabular result now includes those extra columns (generally desirable — it matches NCBI Web BLAST's Description / Scientific Name / Query Cover).

API / IaC diff summary

  • api/services/sharding_precision.py: new enrich_tabular_outfmt() + _PARITY_TABULAR_FIELDS. Expands a bare 6/7 to std first so the shard merge's required evalue + bitscore stay present, then appends any missing parity column.
  • api/routes/elastic_blast.py: the ExternalBlastV1Request validator calls it after the existing shard-merge compatibility check, so both the New Search and Service Bus queue paths enrich.
  • No IaC change. No new dependency.

Why it is safe across the pipeline

  • Shard merge (terminal/merge-sharded-results.sh) resolves its group/rank columns BY NAME and re-emits the full row, so trailing columns (the injected parity fields) are preserved and the # Fields: header lists them.
  • Analytics (api/services/blast/result_analytics.py) map columns by name, so the injected stitle / sscinames / staxids / qcovs populate the Description / Scientific name / Query Cover fields.

Validation evidence

  • uv run ruff check api/services/sharding_precision.py api/routes/elastic_blast.py — clean.
  • uv run pytest -q api/tests/test_sharding_precision.py — 53 passed, including the new enrich tests (bare/std expansion, column preservation, idempotency, XML no-op, merge-compatibility).
  • uv run pytest -q api/tests/test_servicebus_v1_multitoken.py api/tests/test_settings_service_bus.py -k send_preserves api/tests/test_sharded_merge.py api/tests/test_blast_config_sharding.py api/tests/test_web_blast_parity_fixtures.py — green (multitoken / Playground-send / merge / parity assertions updated to the enriched outfmt).
  • End-to-end through the model: ExternalBlastV1Request(... outfmt="7") → blast_options.outfmt == "7 std staxids sscinames stitle qcovs"; a caller's "7 qseqid sseqid pident evalue bitscore" → that list + the parity columns.
  • Full result-page rendering for a live outfmt 7 run is confirmed after the api redeploy.