Description
This workflow is used for gene, compound, drug, and clinical variant queries. The public MCP tools have been consolidated into a new set of NCBI extended interfaces: for genes use search_gene / get_gene_details / get_gene_literature, for compounds use search_compound / get_compound_details / get_compound_literature, and for variants use search_clinvar。
Trigger Conditions
- "What is the function of this gene?"
- "What is the structure or mechanism of this drug?"
- "What is the clinical significance of this variant?"
- Mentions BRCA1, TP53, CYP2C9, warfarin, PubChem, ClinVar
Tool Reference
| Topic | Tool |
|---|---|
| Gene Search | search_gene, get_gene_details, get_gene_literature |
| Compound Search | search_compound, get_compound_details, get_compound_literature |
| Variant / ClinVar | search_clinvar |
| Follow-up Literature Tracking | fetch_article_details, unified_search |
Gene Workflow
1. Search for genes
search_gene(query="BRCA1", organism="human")
2. Get gene details
get_gene_details(gene_id="672")
3. Get gene-related literature
get_gene_literature(gene_id="672", limit=20)
4. Expand PubMed details
fetch_article_details(pmids="12345678,87654321")
get_gene_literatureuses NCBI curated gene-to-publication links, which is usually more accurate than a simple keyword search.
Compound / Drug Workflow
1. Search for compounds
search_compound(query="remimazolam")
2. Get compound details
get_compound_details(cid="11526795")
3. Get compound-related literature
get_compound_literature(cid="11526795", limit=20)
4. Do broader mechanism or comparison searches
unified_search(
query="remimazolam mechanism action",
limit=20,
ranking="quality"
)
ClinVar / Variant Workflow
Search for variants
search_clinvar(query="BRCA1 pathogenic")
search_clinvar(query="NM_007294.4:c.5266dupC")
search_clinvar(query="CYP2C9 drug response")
Key Usage Points
- For gene-level clinical significance:
search_clinvar(query="BRCA1 pathogenic") - For precise variants: use HGVS, for example
NM_007294.4:c.5266dupC - For pharmacogenomics: use gene + phenotype, for example
CYP2C9 drug response
Currently the public interface has no standalone ClinVar detail tool. For greater precision, make the
search_clinvarquery more specific.
Common Integration Workflows
Scenario 1: Gene to Literature
search_gene(query="TP53", organism="human")
get_gene_details(gene_id="7157")
get_gene_literature(gene_id="7157", limit=20)
Scenario 2: Drug to Literature
search_compound(query="propofol")
get_compound_details(cid="4943")
get_compound_literature(cid="4943", limit=20)
Scenario 3: Variant to Mechanism and Literature
search_clinvar(query="BRCA1 pathogenic")
search_gene(query="BRCA1", organism="human")
unified_search(
query="BRCA1 pathogenic variant DNA repair",
limit=20,
ranking="quality"
)
Scenario 4: Pharmacogenomics
search_compound(query="warfarin")
search_gene(query="CYP2C9 warfarin metabolism", organism="human")
search_clinvar(query="CYP2C9 drug response")
unified_search(
query="warfarin CYP2C9 pharmacogenomics",
limit=30,
ranking="quality"
)
Practical Recommendations
Gene
- Prefer official symbols, for example
TP53, rather than vague aliases first - Often supplement with
organism="human", to avoid mixing in mouse or other species
Compound
- Prefer generic names, rather than brand names first
- Once you have the CID, using the CID for everything afterward is the most stable
ClinVar
- The most precise is HGVS nomenclature
- The next best is
gene + pathogenicorgene + condition
Final Principles
The new public tool names should be unified as:
search_geneget_gene_detailsget_gene_literaturesearch_compoundget_compound_detailsget_compound_literaturesearch_clinvar