Bioinformatics Databases

English · 简体中文

Bioinformatics Databases

Wisp bundles 24 MCP servers covering roughly 80 public biology and chemistry databases. You do not configure endpoints or read API docs — you ask a question and the agent picks the database.

"What's the UniProt annotation for P53_HUMAN, and are there structures for it?" → mcp-protein-annotation + mcp-structures-interactions

"Any active phase-3 trials for pembrolizumab in NSCLC?" → mcp-clinical-trials

"gnomAD frequency and CADD score for this variant" → mcp-variants

These ship with the installer — no source tree, no manual install. They live in mcp-servers/bio-tools/.

---

Servers by domain

Literature

Server	Sources
mcp-pubmed	PubMed search and fetch
mcp-literature	OpenAlex, arXiv
mcp-biorxiv	bioRxiv preprints

Also reachable: Europe PMC full text.

Genes, genomes and ontologies

Server	Sources
mcp-genomes	Ensembl REST, UCSC tracks
mcp-genes-ontologies	MyGene, GO / QuickGO, OLS ontologies
mcp-biomart	BioMart queries
mcp-rna	Rfam families

Variants and human genetics

Server	Sources
mcp-variants	gnomAD, CADD
mcp-human-genetics	GWAS Catalog, eQTL Catalogue, PheWeb
mcp-clinical-genomics	ClinVar, dbSNP, ClinGen, CIViC

Expression and regulation

Server	Sources
mcp-expression	GTEx, Human Protein Atlas, expression atlases
mcp-regulation	ENCODE, JASPAR, UniBind
mcp-cellguide	CellxGene cell-type reference
mcp-omics-archives	GEO, ArrayExpress, PRIDE, MetaboLights, MGnify

Proteins, structures and interactions

Server	Sources
mcp-protein-annotation	UniProt, InterPro domains
mcp-structures-interactions	PDB, AlphaFold DB, EMDB, IntAct, STRING, Complex Portal

Chemistry and drugs

Server	Sources
mcp-chemistry	PubChem, ChEBI, Rhea, BindingDB
mcp-chembl	ChEMBL bioactivity, targets, drug search
mcp-zinc	ZINC22 purchasable chemical space (CartBlanche22)
mcp-drug-regulatory	openFDA labels and Drugs@FDA

Clinical and translational

Server	Sources
mcp-clinical-trials	ClinicalTrials.gov, including ESSIE query syntax
mcp-cancer-models	Sanger Cell Model Passports, DepMap, cBioPortal
mcp-research-resources	Antibody Registry, Grants.gov, KEGG, Reactome

Aggregate

Server	What it is
mcp-bio	One aggregated retrieval server spanning 22 domains and 194 tools — the broad fallback when a question does not map cleanly onto a specific server

---

A second bundled server

ketcher-chemistry provides a Ketcher chemical structure editor as an interactive widget — draw or edit a molecule in the conversation rather than describing it in SMILES.

---

Adding your own

Wisp is a standard MCP client (stdio JSON-RPC). Any MCP server you already use works here. Add it in Settings → Connections, where each server's tools can be set to allow / ask / deny individually — a server is not all-or-nothing.

What to keep in mind

• Queries leave your machine. These are public databases; the query and its terms go to the provider. That is the point of them, but it is worth knowing when the query itself is sensitive — an unpublished target name, a patient variant.
• Rate limits are real. Some sources (NCBI in particular) are much faster with an API key. Set those in Settings → Credentials; they are stored in the OS keyring.
• Retrieval is not verification. A record's presence in a database is not peer review. For claims that need it, the bear-* skills go after the actual literature.