Presentation Information
[P02-298]Development of an automated colorimetric measurement system for deep well plates that supports time-series measurements.
○Masahiro Murata1 ,Miho Kitano1 ,Hidenobu Hirayama1,Naoki Watanabe 1 ,Kenta Katayama1,Jun Ishii 1 (1. Kobe Univ. (Japan))
Keywords:
Robotic automation,High-throughput screening
Objective
This study aims to develop a semi-automated system for time-series imaging and quantitative analysis of yeast single-gene knockout library strains in 96-well plates, enabling rapid identification of high carotenoid-producing strains. The system also supports prediction of strain performance and metabolic pathways by integrating image-derived phenotypic data with gene disruption information.
Methods
The system integrates robotic plate handling, imaging, and computational analysis. A robotic arm transfers plates between an incubation pool and an imaging device, enabling continuous bottom-view imaging. The dataset consists of yeast single-gene knockout library strains. Images are processed using OpenCV to detect well positions and extract features such as cell area and color intensity. To address background light interference in low-growth wells, a regression-based colorimetric model was developed to estimate corrected hue values. Time-series data were used to estimate growth rates and carotenoid production, enabling prediction of optical density (OD), pigment accumulation, and contamination. Outputs from the color model were combined with gene disruption data to manually infer high-producing strains and associated metabolic pathways.
Results
The system significantly reduced manual workload in data acquisition and analysis. Hue measurement time decreased to approximately one-twentieth of conventional methods. The regression model improved robustness of carotenoid estimation across growth conditions. Integration of image features with gene information enabled effective identification of high-producing strains and provided insights into metabolic pathways. The predictive framework also allowed early detection of contamination and promising strains.
Discussion
Combining automation, image-based phenotyping, and gene-level data improved both throughput and interpretability. The color correction model enhanced reliability under variable growth conditions. Integration of genotype and phenotype enabled biologically meaningful insights. Although pathway prediction was performed manually, future automation could improve scalability and objectivity.
Conclusion
This system provides a scalable platform for high-throughput, time-resolved screening of yeast knockout strains. By integrating automation, image analysis, and gene-informed interpretation, it enables efficient identification of high carotenoid-producing strains and contributes to understanding genotype–phenotype relationships.
This study aims to develop a semi-automated system for time-series imaging and quantitative analysis of yeast single-gene knockout library strains in 96-well plates, enabling rapid identification of high carotenoid-producing strains. The system also supports prediction of strain performance and metabolic pathways by integrating image-derived phenotypic data with gene disruption information.
Methods
The system integrates robotic plate handling, imaging, and computational analysis. A robotic arm transfers plates between an incubation pool and an imaging device, enabling continuous bottom-view imaging. The dataset consists of yeast single-gene knockout library strains. Images are processed using OpenCV to detect well positions and extract features such as cell area and color intensity. To address background light interference in low-growth wells, a regression-based colorimetric model was developed to estimate corrected hue values. Time-series data were used to estimate growth rates and carotenoid production, enabling prediction of optical density (OD), pigment accumulation, and contamination. Outputs from the color model were combined with gene disruption data to manually infer high-producing strains and associated metabolic pathways.
Results
The system significantly reduced manual workload in data acquisition and analysis. Hue measurement time decreased to approximately one-twentieth of conventional methods. The regression model improved robustness of carotenoid estimation across growth conditions. Integration of image features with gene information enabled effective identification of high-producing strains and provided insights into metabolic pathways. The predictive framework also allowed early detection of contamination and promising strains.
Discussion
Combining automation, image-based phenotyping, and gene-level data improved both throughput and interpretability. The color correction model enhanced reliability under variable growth conditions. Integration of genotype and phenotype enabled biologically meaningful insights. Although pathway prediction was performed manually, future automation could improve scalability and objectivity.
Conclusion
This system provides a scalable platform for high-throughput, time-resolved screening of yeast knockout strains. By integrating automation, image analysis, and gene-informed interpretation, it enables efficient identification of high carotenoid-producing strains and contributes to understanding genotype–phenotype relationships.
Comment
To browse or post comments, you must log in.Log in