search_genesLABSCO SUMMARY
Alliance data and AllianceMine, from one server.
The API tools cover what you would otherwise click through alliancegenome.org for — what a gene is linked to, its orthologs in mouse, where it is expressed. AllianceMine is the reason to keep it: when the fixed lookups do not answer the question, you can write the query yourself.
What it is
A Node server over the Alliance of Genome Resources. Eleven tools cover the Alliance API — genes, diseases, expression, orthologs, phenotypes, interactions and alleles across human, mouse, rat, zebrafish, fly, worm, yeast and xenopus — and eleven more open AllianceMine, the query engine behind it.
What you get
- Genes searched with an optional species filter, and one gene read in full — symbol, location, synonyms (
search_genes,get_gene_info) - Disease associations for a gene, and disease search by name (
get_gene_diseases,search_diseases) - Expression across tissues and stages, phenotype annotations, molecular and genetic interactions (
get_gene_expression,get_gene_phenotypes,get_gene_interactions) - Cross-species homologs, the alleles and variants recorded for a gene, allele search, and the list of supported organisms (
find_orthologs,get_gene_alleles,search_alleles,get_species_list) - AllianceMine reached three ways — raw PathQuery XML, a JSON query builder with from/select/where, and a natural-language mode that hands the schema back for the model to work from (
mine_query,mine_query_builder,mine_natural_query,mine_search) - Prebuilt AllianceMine templates such as
Gene_Alleles,Gene_DOTerm,Gene_GOTermsandGOTerm_Genes, listed and run (mine_list_templates,mine_run_template) - Gene and protein lists fetched, created, added to and deleted (
mine_get_lists,mine_get_list,mine_create_list,mine_add_to_list,mine_delete_list)
Requirements
npx -y agr-mcp-server in the client config, or a global npm install. No key and no account for the lookups — they run against the Alliance's public API and AllianceMine. Creating, adding to or deleting a list needs an AllianceMine token in ALLIANCEMINE_TOKEN. Gene identifiers are the source-native ones and are case-sensitive: HGNC:1100 for human, MGI:95892 for mouse, WB:WBGene00000898 for worm, and so on.
Setup effort
One command — npm install -g agr-mcp-server
USE CASES
01
Finding out what a gene is implicated in
Search by symbol, take the identifier, and read the disease associations and phenotype annotations recorded against it. This is the first pass on a gene that came out of someone else's screen.
02
Moving a finding between organisms
Take the human identifier and pull its orthologs, then run the same disease and expression lookups against the mouse or zebrafish gene. The identifier formats differ per species, and the ortholog call is what bridges them.
03
Checking where and when a gene is expressed
Ask for the expression data across tissues and stages before designing anything around it. A gene that is only on in one stage changes what an experiment has to look like.
04
Asking something the fixed lookups do not cover
Drop into AllianceMine. Describe the query in JSON — what to select, what to filter on — or hand over raw PathQuery XML, or ask in words and get the schema back to build against.
05
Working through a gene set over several sessions
Keep the set as an AllianceMine list rather than pasting identifiers each time: create it, add to it as the analysis grows, and read it back next session. List writes are the one part that needs a token.
06
Running a standard query without writing one
List the AllianceMine templates and run the one that matches — alleles for a gene, disease annotations, GO terms, or the genes behind a GO term. Faster than assembling the same query by hand.
22 tools
Listed in project docs. Your client may show a different list.
get_gene_infoget_gene_diseasessearch_diseasesget_gene_expressionfind_orthologsget_gene_phenotypesget_gene_interactionsget_gene_allelessearch_allelesget_species_listmine_searchmine_querymine_query_buildermine_natural_querymine_list_templatesmine_run_templatemine_get_listsmine_get_listmine_create_listmine_add_to_listmine_delete_listBEFORE YOU INSTALL
What to know first
Tools you get
22 tools named in the project's documentation.
Where it runs
All 4 documented clients run the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
npm install -g agr-mcp-serverMCP Clients
Claude Code, Cursor, Windsurf, Claude Desktop
JSON
{
"mcpServers": {
"agr-genomics": {
"command": "npx",
"args": [
"-y",
"agr-mcp-server"
]
}
}
}
