Disease¶
Use disease commands for normalization and disease-centric cross-entity pivots.
For cancer-outcome questions, the disease survival section adds SEER Explorer
survival context without creating a separate survival entity.
Search diseases¶
biomcp search disease -q melanoma --limit 5
biomcp search disease -q glioblastoma --source mondo --limit 5
Search resolves common labels toward canonical ontology-backed identifiers.
Get disease records¶
By label:
By MONDO identifier:
The base disease card includes concise OpenTargets gene-score summaries when OpenTargets
returns ranked associated targets. Prefer canonical MONDO:<id> values in automation:
they are the stable form BioMCP uses for normalization and fallback repair.
When ranked disease-gene context is present, the See also: block also promotes
the strongest follow-up gene pivot before the generic disease-level searches,
for example biomcp get gene SCN1A clingen constraint on a Dravet syndrome
gene card.
The default disease card's More: block keeps genes, pathways, and
phenotypes visible while also surfacing survival and funding so those
opt-in sections stay discoverable from the base card.
Disease sections¶
Genes (Monarch-backed rows plus additive CIViC and OpenTargets disease-gene associations; OpenTargets scores attach to any rendered row with a matching target score):
Phenotypes (compact Key Features summary plus the comprehensive HPO annotation list):
When BioMCP can extract a reliable disease summary, the phenotype section renders
### Key Features above the HPO table. That summary is also exposed as
key_features[] in --json output. The table remains the comprehensive phenotype
annotation list, and the existing completeness note still applies.
Clinical features (MedlinePlus clinical-summary rows):
This opt-in section is currently available for the reviewed configured disease
set: uterine leiomyoma / uterine fibroid, endometriosis, and chronic venous
insufficiency. Rows are source-native MedlinePlus clinical-summary statements
with auditable HPO mapping metadata. Unsupported diseases omit the
clinical_features field rather than fabricating rows, and the section is not
included in biomcp get disease <name_or_id> all.
Variants (CIViC disease-associated variants):
When the variants section is loaded, JSON also exposes top_variant as the
highest-ranked CIViC-backed association, and markdown shows the same compact
anchor above the full variants table.
Models (Monarch model-organism evidence):
Pathways (associated pathways):
Prevalence (prevalence data):
Survival (SEER Explorer 5-year relative survival by sex for mapped cancers):
The survival section is filtered to all ages and all races / ethnicities. When
the normalized disease does not map cleanly to one SEER cancer site, BioMCP
returns a stable survival_note instead of failing the disease card.
Diagnostic-test pivot (GTR and WHO IVD tests for the condition):
The disease diagnostic card is capped at 10 rows so it stays terminal-sized.
When rows exist, BioMCP prints a See also: command such as
biomcp search diagnostic --disease tuberculosis --source all --limit 50 for
the broader paged diagnostic search; use --offset on search diagnostic to
continue paging.
Funding (NIH Reporter grants for the requested disease phrase, with canonical-name fallback for identifier lookups, over the most recent 5 NIH fiscal years):
The diagnostics, DisGeNET, funding, and clinical features sections stay opt-in
and are not included in biomcp get disease <name_or_id> all.
CIViC (clinical evidence):
Combined sections:
biomcp get disease MONDO:0005105 genes phenotypes variants models
biomcp get disease "Marfan syndrome" funding
biomcp get disease "chronic myeloid leukemia" survival
biomcp get disease MONDO:0005105 all
Helper commands¶
biomcp disease trials melanoma --limit 5
biomcp disease drugs melanoma --limit 5
biomcp disease articles "Lynch syndrome" --limit 5
Phenotype-to-disease search¶
Use HPO term sets for ranked disease candidates:
You can pass terms space-separated or comma-separated.
Typical disease-centric workflow¶
- Normalize disease label.
- Pull disease sections (
genes,phenotypes,variants,models, andsurvivalfor cancers) for context. - Use normalized concept for trial or article searches.
Example:
biomcp get disease MONDO:0005105 genes phenotypes
biomcp get disease "chronic myeloid leukemia" survival
biomcp search trial -c melanoma --status recruiting --limit 5
biomcp search article -d melanoma --limit 5
JSON mode¶
biomcp --json get disease MONDO:0005105 includes top_gene_scores[] with
overall OpenTargets scores and any available GWAS, rare-variant, or somatic subtype scores.
Practical tips¶
- Prefer MONDO IDs in automation workflows.
- Keep raw labels in user-facing notes for readability.
- Pair disease normalization with biomarker filters for trial matching.