Skip to content

External BLAST API evalue default aligned to NCBI (0.05)

Motivation

Result parity with NCBI Web BLAST depends on the request-side options matching NCBI's defaults. The dashboard submit form already defaulted the expect-value (evalue) threshold to 0.05 — the NCBI Web BLAST megablast default — and its preset table labels 0.05 as "Standard (default)". The external OpenAPI submit facade (POST /api/v1/elastic-blast/submit), however, still defaulted options.evalue to 10, so a programmatic caller who omitted evalue got a much looser threshold than an equivalent UI submission, producing a longer, non-NCBI-equivalent hit tail.

User-facing change

ExternalBlastOptions.evalue now defaults to 0.05 instead of 10. Callers of /api/v1/elastic-blast/submit that omit options.evalue now get the same threshold as the dashboard form and NCBI Web BLAST. Callers that pass an explicit evalue are unaffected. The field validation (gt=0) is unchanged.

API / IaC diff summary

  • api/routes/elastic_blast.py — ExternalBlastOptions.evalue default 10.0 → 0.05, with a description noting the NCBI / dashboard alignment.

No other endpoint changed. The result-analytics post-filter default (max_evalue=10.0 in api/routes/blast/result_analytics.py) is a separate display filter on already-computed hits and is intentionally left unchanged.

Validation evidence

  • uv run pytest -q api/tests/test_external_blast_api.py — new test_external_blast_options_default_evalue_matches_ncbi asserts an omitted options block yields evalue == 0.05; existing forwarding tests (which pass evalue explicitly) stay green.
  • uv run pytest -q api/tests/test_blast_submit_route_options.py — UI/OpenAPI shared-execution-config tests green.
  • uv run pytest -q api/tests — full suite green (one unrelated test_terminal_exec parallel-timing flake passed on isolated re-run).
  • uv run ruff check api/routes/elastic_blast.py.