TimeFlow: A Density-Driven Pseudotime Method for Flow Cytometry Data Analysis.

Area of Science:

• Computational Biology
• Single-cell Analysis
• Immunology

Background:

• Pseudotime methods are crucial for ordering cells during differentiation.
• Existing methods may face challenges with complex, multi-dimensional datasets.

Purpose of the Study:

• To introduce TimeFlow, a new computational method for pseudotime ordering in multi-dimensional flow cytometry data.
• To assess TimeFlow's performance and generalizability across different cell types and patient samples.

Main Methods:

• Developed TimeFlow, utilizing a normalizing flow model to estimate cell population density and track differentiation paths on a graph.
• Applied TimeFlow to multi-dimensional flow cytometry datasets from human bone marrow samples, encompassing various hematopoietic cell lineages.

Main Results:

• TimeFlow successfully computed fine-grained pseudotime, aligning with known human hematopoiesis.
• The method demonstrated strong performance, generalizing across patients and unseen cell states, outperforming 11 other pseudotime methods.
• Results showed utility in modeling cell surface protein dynamics and potential for automated cell lineage detection.

Conclusions:

• TimeFlow offers a robust and accurate approach for pseudotime analysis in complex flow cytometry data.
• The method provides valuable insights into cellular differentiation dynamics and holds promise for future applications in lineage tracing.