BioThings MCP

BioThings MCP wraps three public bioinformatics APIs—mygene.info, myvariant.info, and mychem.info—into a single MCP server. You get gene annotations, variant lookups, and chemical compound data without writing your own HTTP clients or wrestling with endpoint documentation. I'd reach for this when I'm prototyping a bioinformatics pipeline in Claude and need to cross-reference gene symbols, check variant pathogenicity, or pull compound structures without leaving the conversation. The practical win is speed. Instead of context-switching to NCBI or PubChem, you ask Claude to look up BRCA1 or rs7412 and get structured JSON back immediately. The server queries the BioThings APIs under the hood, so you're not inventing a data source—you're just making it conversational. This matters when you're iterating on a hypothesis or building a quick proof-of-concept for a research collaborator. Quirks: the server is as good as the upstream APIs. If mygene.info is slow or returns partial data, you'll see that here. The MCP doesn't cache or retry, so you're at the mercy of network conditions. Also, the tool list is small—three APIs, three query types. If you need PDB structures or pathway data, you'll need another server. The documentation is minimal, so you'll spend a few minutes poking around to understand what each tool actually returns. Who shouldn't bother: if you're not working in genomics, drug discovery, or bioinformatics, this is too niche. If you already have a polished API client or a local database, the MCP adds little. But if you're a computational biologist who wants to ask Claude 'what does this variant do' without writing a script, it's a clean, no-config shortcut. The install is one command, no API keys, no setup friction. That's rare enough to be worth it.