Skip to content

Phenotype

Use phenotype commands to resolve HPO IDs and labels, rank semantic-similarity candidates through Monarch, and check the returned page for exact direct phenotype associations. Similarity is not evidence that a disease has the requested phenotype.

Search phenotypes

By HPO identifiers (space- or comma-separated):

biomcp search phenotype "HP:0001250 HP:0001263"
biomcp search phenotype "HP:0001250,HP:0001263"

By one symptom phrase:

biomcp search phenotype "developmental delay"

By multiple symptom phrases (comma-separated):

biomcp search phenotype "seizure, developmental delay"

Multiple terms with limit:

biomcp search phenotype "HP:0001250 HP:0001263" --limit 20

The positional terms argument accepts:

  • canonical HPO IDs, space- or comma-separated
  • one symptom phrase
  • multiple symptom phrases separated by commas

Free-text symptom phrases are resolved to HPO IDs before the Monarch similarity search runs. Every phrase must resolve; BioMCP does not silently discard a phrase with no HPO match. A query may contain at most 10 phrases and at most 10 unique resolved HPO terms. Output records the original phrase, normalized HPO ID, and HPO label for every resolved term. Use --limit and --offset within Monarch's first 50 ranked matches; offset + limit cannot exceed 50. When that provider window is exhausted, BioMCP reports possible truncation and asks you to refine the terms instead of emitting an unusable continuation.

Get records

Phenotype is search-only. There is no get phenotype subcommand.

Request sections

Each candidate retains its semantic similarity score and reports one exact direct-support state per resolved HPO term:

  • supported: an exact, positive direct Monarch disease-to-phenotype row exists
  • not_supported: a complete lookup contains no such row; this does not prove the disease lacks the phenotype
  • indeterminate: truncation, missing fields, or inconsistent rows prevent a claim of absence
  • unavailable: direct-support enrichment failed or exceeded its deadline; similarity results remain usable as candidates

Phenotype search rows do not expose extra section names. Use search disease or get disease <id> phenotypes when you want a normalized disease follow-up.

Helper commands

Phenotype is search-only. Start with search phenotype for HPO term sets or symptom phrases, then switch to disease commands once you have the right normalized concept. If you want to inspect candidate HPO terms first, run biomcp discover "<symptom text>" and use the suggested HP: IDs. Markdown and JSON suggest biomcp get disease <MONDO_ID> phenotypes only for the first provider-ordered candidate that is supported for every resolved term. If no row meets that rule, the disease follow-up is suppressed rather than falling back to the first similarity result. Pagination remains the first continuation when present, and biomcp list phenotype remains available.

JSON mode

biomcp --json search phenotype "HP:0001250"

Practical tips

  • Use HPO IDs for precise lookups when you know the exact term.
  • Use commas to separate multiple symptom phrases in one search.
  • Combine multiple HPO IDs in a single query to retrieve a phenotype set.
  • Prefer 2-5 high-confidence HPO IDs when you already know them.