High-speed cell partitioning through reactive machine learning-guided inkjet printing.

Area of Science:

• Genomics
• Biotechnology
• Cell Biology

Background:

• Single cell partitioning in nanowells is crucial for molecular assays but limited by slow cell dispensing speeds.
• Achieving sufficient statistical power in single cell experiments requires rapid cell isolation.
• Existing methods face challenges in balancing speed and accuracy for high-throughput applications.

Purpose of the Study:

• To develop a high-throughput single cell dispensing instrument for rapid cell partitioning into open substrates.
• To enhance the speed of single cell isolation for genomic assays, such as direct-transposition single cell whole genome sequencing (scWGS).
• To improve the efficiency and scalability of single cell analysis workflows.

Main Methods:

• Development of the Isolatrix instrument utilizing inkjet technology with real-time optical feedback.
• Integration of machine learning algorithms for real-time classification of single cell events post-dispensing.
• Training the machine learning classifier on manually labeled data and validation using fluorescent imaging and genomic analysis.

Main Results:

• The Isolatrix achieved up to a 9.69 times increase in isolation speed compared to existing predictive workflows.
• Demonstrated high classification accuracy of 98.7% at a rate of 0.52 seconds per single cell.
• Generated scWGS profiles with low background contamination, high genomic coverage uniformity, and enabled detection of copy number alterations.

Conclusions:

• The Isolatrix instrument significantly accelerates high-throughput single cell isolation for genomic applications.
• Machine learning-driven, real-time classification enhances isolation speed and accuracy in single cell dispensing.
• This technology is poised to enable large-scale genomic profiling across various data modalities.