PhenoComb: a discovery tool to assess complex phenotypes in high-dimensional single-cell datasets.

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• High-dimensional cytometry enables detailed cell analysis.
• Manual gating methods are limited by prior biological knowledge, potentially missing novel phenotypes.
• Agnostic exploration of all marker combinations is needed for comprehensive analysis.

Purpose of the Study:

• To introduce PhenoComb, an R package for agnostic phenotype exploration in high-dimensional cytometry data.
• To enable the assessment of all possible marker combinations for unbiased discovery.
• To provide tools for statistical comparison and relevance evaluation of identified phenotypes.

Main Methods:

• PhenoComb utilizes signal intensity thresholds to discretize marker states.
• It counts cells across all possible marker combinations in a memory-efficient manner.
• The package offers adjustable parameters for guided analysis, including population filtering and complexity definition.

Main Results:

• PhenoComb efficiently processes datasets with numerous markers (e.g., 26 markers in hours).
• Applied to HIV and COVID-19 datasets, it identified relevant immune phenotypes associated with disease states.
• The tool successfully identified known and novel immune phenotypes in public datasets.

Conclusions:

• PhenoComb provides a powerful, agnostic approach for discovering biologically relevant phenotypes in high-dimensional single-cell data.
• It overcomes the limitations of manual gating by exploring all marker combinations.
• The R package is a valuable tool for researchers in immunology and computational biology.