Thumbnails:
List:
Year:
Category:
Session:
Poster:
Getting poster data...
Julia Patricia Schessner, Mikhail Lebedev, Magnus Schwörer, Patricia Skowronek, Matthias Mann (Max-Planck-Institute for Biochemistry, Martinsried, Germany)
Proteomics technology has become increasingly accessible to researchers across disciplines, yet the complex data analysis presents a significant challenge for non-specialists and is often the bottleneck. More recently, large language models (LLMs) have transformed all aspects of science, from writing proposals, code and papers, to planning and analyzing experiments. This poster shows, how we combined the interactive proteomics data analysis tool AlphaPeptStats with an interface to LLMs to develop the Proteome Interpreter (PI): An analysis co-pilot, with which we aim to bridge knowledge gaps in an increasingly interdisciplinary scientific community. By equipping the PI with tools to run statistical analyses, slice and visualize data and communicate with external APIs, previously complex analysis steps can now quickly and easily be addressed using natural language. Through tooling and prompt engineering we mitigate issues of current LLMs with hallucinations and reproducibility. Ultimately, the PI helps with formulating hypotheses and follow-up experiments. Importantly, the researcher stays in charge of the analysis, but is no longer slowed down by knowledge gaps or software barriers.