cytometree: A binary tree algorithm for automatic gating in cytometry analysis

Area of Science:

• Biotechnology
• Computational Biology
• Immunology

Background:

• Flow cytometry is essential for single-cell protein quantification and immunophenotyping.
• Manual analysis of complex flow cytometry data is subjective and time-consuming.
• Automated methods are needed to handle increasing experimental complexity.

Purpose of the Study:

• To introduce cytometree, a novel unsupervised algorithm for automated cell population identification (gating) in flow cytometry.
• To provide an objective and efficient alternative to manual data analysis.

Main Methods:

• Cytometree constructs a binary tree where nodes represent cell subpopulations.
• Marker distributions at each node are modeled using mixtures of normal distributions.
• A model selection procedure based on Akaike information criteria guides node splitting.

Main Results:

• Cytometree demonstrated superior performance compared to state-of-the-art unsupervised algorithms on benchmark datasets.
• The algorithm outperformed the best open-source alternative on a Human Immunology Project Consortium T-cell panel.
• Cytometree achieved the shortest computation time among tested unsupervised algorithms.

Conclusions:

• Cytometree offers an accurate and efficient solution for automated flow cytometry data analysis.
• The algorithm effectively addresses the challenges posed by complex, high-dimensional cytometry experiments.
• Cytometree represents a significant advancement in computational cytometry for immunological research.