Genomics MCP — Claude Code MCP server: data access and setup
Genomics MCP is a Claude Code MCP server for retrieving genomic datasets and reference records. An agent can request a specific locus from an archive or an indexed file.
MIT ★ 5 updated 2026-11-17
claude mcp add genomics-mcp -- docker run -i --rm ghcr.io/rewire-bio/genomics-mcp:0.1.0What Genomics MCP does
Genomics MCP retrieves genomic regions and attaches source details to its results. Its tools cover dataset searches, sample comparisons and reference lookups. It supports archives such as EGA and ENCODE alongside indexed files. Clinical interpretation and variant calling are outside its scope.
Region requests must name the genome assembly and use coordinates counted from 0, excluding the end boundary. Assemblies are not converted. Partial reads depend on format and server support; downloading an entire file is a separate action with a transfer budget. Local access is restricted to permitted folders.
The container needs Docker and is tested on Linux x86_64. Building from source requires Python 3.12, uv, a C compiler, plus development files for libcurl and zlib; pyBigWig is compiled with remote access support. AlphaGenome Atlas needs a personal key. EGA test access uses GENOMICS_MCP_EGA_PUBLIC_TEST_ACCOUNT=1. The unreleased SRA-to-FASTQ conversion requires SRA Toolkit. Local Streamable HTTP connections require a bearer token.
What it helps with
- Locate public genomic datasets and their files
- Extract a selected region from BAM or VCF
- Compare genotypes between samples
- Retrieve ClinVar entries for a variant
- Trace reference results to their sources and versions
Example prompts for the agent
- Ask Genomics MCP for ClinVar entries associated with
7-140753336-A-Ton GRCh38, separating the classification categories. - With Genomics MCP, calculate the average signal for
ENCFF792QDSusing assembly GRCh38 and intervalchr1:[1000000, 1001000). - Have Genomics MCP locate ENA accession
DQ285577.1and download its sequence as FASTA.
How to install Genomics MCP
Platform CLI
claude mcp add genomics-mcp -- docker run -i --rm ghcr.io/rewire-bio/genomics-mcp:0.1.0Platform CLI
codex mcp add genomics-mcp -- docker run -i --rm ghcr.io/rewire-bio/genomics-mcp:0.1.0Config file
{
"mcpServers": {
"genomics-mcp": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/rewire-bio/genomics-mcp:0.1.0"
]
}
}
}Config file
{
"servers": {
"genomics-mcp": {
"type": "stdio",
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/rewire-bio/genomics-mcp:0.1.0"
]
}
}
}Platform CLI
gemini mcp add genomics-mcp docker run -i --rm ghcr.io/rewire-bio/genomics-mcp:0.1.0Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Genomics MCP
How do I install Genomics MCP in Claude Code?
The container setup requires Docker. Register the server with claude mcp add genomics-mcp -- docker run -i --rm ghcr.io/rewire-bio/genomics-mcp:0.1.0.
What does Genomics MCP do?
It lets an agent search genomic archives, retrieve regions from indexed files and query reference databases. Results include source tracking and access status; the server does not infer biological conclusions.
What does Genomics MCP need to run?
Docker is required for the container route; a source build needs Python 3.12, uv, a C compiler, and development files for libcurl and zlib. Optional AlphaGenome Atlas access needs a personal key, EGA test access uses GENOMICS_MCP_EGA_PUBLIC_TEST_ACCOUNT=1, and the unreleased SRA-to-FASTQ feature depends on SRA Toolkit. Local Streamable HTTP requires a bearer token, while local files must be inside permitted folders.