A Setup for Automatic Raman Measurements in High-Throughput Experimentation.

Area of Science:

• Biotechnology and Analytical Chemistry
• Process Engineering and Automation
• Machine Learning in Scientific Research

Background:

• High-throughput (HT) experimentation accelerates biological research but is often limited by slow analytical methods.
• Current analytical techniques struggle to keep pace with the rapid sample generation in HT workflows.
• There is a need for automated, high-speed analytical solutions to fully leverage HT capabilities.

Purpose of the Study:

• To develop and validate an integrated system for automated, high-throughput Raman spectral measurements.
• To accelerate the analysis of metabolite concentrations in biological samples.
• To enable the generation of large datasets for machine learning model training in biotechnology.

Main Methods:

• Development of an automated system integrating physical devices and software for Raman spectroscopy.
• Simultaneous handling and measurement of eight parallel 50 μL samples.
• Implementation of a machine learning model for predicting metabolite concentrations (glucose and acetate) from Raman spectra.

Main Results:

• The system completes measurement, handling, cleaning, and concentration prediction within 45 seconds per sample.
• Machine learning model achieved mean absolute errors of 0.27 g L⁻¹ for glucose and 0.06 g L⁻¹ for acetate.
• Demonstrated consistent high-throughput spectral data collection for fermentation monitoring.

Conclusions:

• The automated Raman spectral measurement system significantly accelerates analytical throughput in biotechnology.
• The integrated machine learning approach enables accurate prediction of metabolite concentrations.
• This technology supports the generation of extensive datasets for developing robust machine learning models for bioprocess analysis.