I have createda standalone MCP (Model-Context Protocol) server that wraps the GraphQL API into a set of simple, easy-to-use tools.
This allows an AI agent to perform complex, multi-step queries without needing to construct GraphQL.
You can check it out on GitHub: GitHub - nickzren/opentargets-mcp: MCP server for Open Targets Data
I hope the community finds it useful and I welcome any feedback or contributions!
It looks amazing! What are your plans for the MCP @nickzren? Are you planning to maintain it? Expand it? Are you looking for collaborators? Can you tell us a bit more?
There are efforts underway within Open Targets to initiate the development of an MCP. It’s still early days. If everything goes well, there will be an official MCP in the future, but it would be great to work with others if they are thinking along the same lines
I’ve been using OT data for a long time, and we normally just download the entire dataset and store it in a relational DB for further use. I wanted to set up a way for a local LLM app, like Claude Desktop, to access OT data directly. This makes it easier to get answers from OT data with simple, human questions.
Yes, I plan to maintain it as I start using it more and get feedback from potential users here. It would be great if others are interested and become collaborators on this repo.
It’s also great to hear that an official MCP is planned for the future, and I’m looking forward to using it.