Nicholas (Bits in Bio founder) over at Sphinx Bio recently released platechain, an open-source Python package for universal plate parsing. The repo includes examples of plate readers from Tecan and Byonoy.
The package allows an operator to convert this…
With minimal effort!
Behind the scenes, platechain uses LangChain for prompt engineering & LLM interactions. One can even deploy platechain as an API with LangServe for a user-friendly interactive interface.
There are some limitations but overall this is a pretty interesting use case for LLM’s in day-to-day lab environments where maybe programming expertise is minimal.
Read the blog post here and also check out their LRIG Bay Area presentation.
Is Langchain local? Or an external service? I’m not sure how I’d feel sending my data to an external service…
You could also swap out the OpenAI model for a self hosted open source model if you wanted to run everything locally! LangChain makes this easy to do (though you’d have the added challenge of hosting an LLM locally).
Does that make sense?
Awesome work, reminds me of the AI assistant for PyHamilton/ PLR
Hey, thanks for letting us know. In general companies I’ve talked to are very hesitant to use OpenAI/ChatGPT with data-related tasks, just due to possible IP related issues, so it’s good to know that we can swap out the model for a self-hosted one. Great project!