Automated cell type discovery and classification through knowledge transfer.

Area of Science:

• Single-cell analysis
• Computational biology
• Immunology

Background:

• Mass cytometry enables high-dimensional, single-cell measurements.
• High-dimensional data presents computational challenges for analysis.
• Manual gating is a bottleneck for large-scale mass cytometry studies.

Purpose of the Study:

• To develop a fully automated algorithm for mass cytometry data analysis.
• To accurately classify canonical and discover novel cell types.
• To improve the reliability and interpretability of mass cytometry results.

Main Methods:

• Developed Automated Cell-type Discovery and Classification (ACDC) algorithm.
• Utilized machine learning for automated cell classification.
• Evaluated ACDC performance on real-world mass cytometry datasets.

Main Results:

• ACDC accurately classifies canonical cell populations.
• ACDC successfully highlights novel cell types.
• ACDC provides reliable estimations comparable to manual gating.
• ACDC automates classification of ambiguous cell types, facilitating discovery.

Conclusions:

• ACDC significantly enhances the reliability of mass cytometry data analysis.
• ACDC improves the interpretability of high-dimensional single-cell profiling.
• ACDC offers a fully automated solution, overcoming limitations of manual inspection.