We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/genomoncology/biomcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
from biomcp.articles.autocomplete import Entity, EntityRequest, autocomplete
async def test_autocomplete(anyio_backend, http_cache):
# new cache for each call
assert http_cache.count == 0
# gene (compare using entity_id directly)
request = EntityRequest(concept="gene", query="her2")
entity = await autocomplete(request=request)
assert entity.entity_id == "@GENE_ERBB2"
# variant
request = EntityRequest(concept="variant", query="BRAF V600E")
assert await autocomplete(request=request) == Entity(
_id="@VARIANT_p.V600E_BRAF_human",
biotype="variant",
name="p.V600E",
)
# disease
request = EntityRequest(concept="disease", query="lung adenocarcinoma")
assert await autocomplete(request=request) == Entity(
_id="@DISEASE_Adenocarcinoma_of_Lung",
biotype="disease",
name="Adenocarcinoma of Lung",
match="Multiple matches",
)
assert http_cache.count == 3
# duplicate request uses the cached response
request = EntityRequest(concept="gene", query="her2")
entity = await autocomplete(request=request)
assert entity.entity_id == "@GENE_ERBB2"
assert http_cache.count == 3