Automated identification of maximal differential cell populations in flow cytometry data

Area of Science:

• Immunology
• Computational Biology
• Biostatistics

Background:

• Flow cytometry is crucial for analyzing cell populations.
• Identifying robust biomarkers from complex flow cytometry data remains challenging.
• Existing methods may not distinguish true disease-associated cell changes from related population shifts.

Purpose of the Study:

• To introduce a novel method for discovering accurate candidate biomarkers from flow cytometry data.
• To define and identify a new class of biomarkers termed 'driver cell populations'.
• To provide an interpretable visualization tool for these biomarkers.

Main Methods:

• Development of a new cell population scoring method, Specific Enrichment (SpecEnr).
• Identification of driver cell populations whose abundance correlates with sample class (e.g., disease) independently of related populations.
• Implementation of a lattice-based visualization tool for interpretability.
• The method is available as the R package flowGraph.

Main Results:

• The SpecEnr score effectively identifies candidate biomarkers.
• The method successfully detects driver cell populations.
• The lattice visualization aids in the interpretation of identified driver cell populations.
• The flowGraph R package provides a practical implementation.

Conclusions:

• The novel method and SpecEnr score offer a robust approach to biomarker discovery in flow cytometry.
• Driver cell populations represent a valuable new class of biomarkers.
• The flowGraph package facilitates the application and interpretation of this method in biological research.