Jupyter Notebook in Lab Automation for Data Efficiency and Transparency — 97a — Landon Johnson, Tim Hartman, Bichar Shrestha Gurung, Etienne Gnimpieba

Automated systems, such as liquid handling robots and data analysis platforms, enable researchers to conduct high-throughput assays with minimal human intervention, reducing potential error and contamination. These ‘experiment-ready’ systems are often expensive and render previously purchased equipment obsolete, hindering their adoption. Chatbots and AI that can fabricate experiments and fake research also bring skepticism about the widespread use of AI in science. Furthermore, electronic lab notebooks and automation schedulers are limited in their ability to design, run, document, and report biology experiments. Herein, we propose a collaborative framework where technicians, AI, robots, and analog wet lab instruments work together to conduct biological experiments. Through orchestration in Jupyter Notebook, we can ensure unified data and protocol management to facilitate cost-effective, transparent, and efficient automation adoption.

Our system’s workflow begins in Jupyter Notebook and Python, with our first use case involving simple E. coli growth analysis. Research questions are drafted by the user with the intent to query a chatbot to design an experiment to address them. This is followed by AI-assisted experimental design and instrument programming. All reagents and labware are logged with optical character recognition to log essential data for quality control. Protocol execution is a collaboration between the technician and high-throughput analysis instruments, with video recordings as proof of execution. All data, images, and videos are pulled into the notebook for future review and reporting by scientists and AI.

This approach leverages an inexpensive, collaborative automation workflow that is friendly to existing lab instruments yet has a view toward AI-assisted robot science. It prioritizes maximum data collection to ensure experiment transparency and FAIR compliance and as a bulwark against fabricated research. Ongoing work involves integrating more AI tools to assist researchers in the design, redesign, execution, analysis, and publication of experimental results.

University of South Dakota
Etienne Gnimpieba