CyTOF workflow

CyTOF workflow analyzes high-dimensional mass and flow cytometry (HDCyto/CyTOF) data in R using Bioconductor packages to define cell populations and perform differential discovery across complex experimental designs.

Key Features:

• R/Bioconductor integration: Implements analyses in R and integrates Bioconductor packages for high-dimensional cytometry workflows.
• Cell Population Definition (FlowSOM clustering): Uses FlowSOM clustering to computationally define cell populations based on phenotypic markers.
• Reproducible Manual Merging: Provides an optional reproducible manual merging step to refine clusters generated by FlowSOM.
• Multiple Analysis Paths: Supports association of cell-type abundance with phenotypic traits, examination of signaling marker changes within specific subpopulations, and differential analyses of aggregated signals across samples.
• Regression-Based Differential Analyses: Applies regression frameworks that treat cytometry measurements as the response to model complex experimental designs.
• Mixed Models and Batch Effects: Accommodates batch effects and paired designs using generalized linear mixed models (GLMM) or linear mixed models (LMM).
• Handling Overdispersion: Models overdispersion in cell counts or aggregated signals within the regression frameworks.
• Exploratory Data Analysis and Visualization: Includes quality control with multi-dimensional scaling plots, cluster reporting via dimensionality reduction and heatmaps with dendrograms, and visualization of differential analyses including plots of aggregated signals.

Scientific Applications:

• Immunophenotyping: Characterization of immune cell populations in clinical and basic research settings.
• Investigation of Cellular Signaling: Analysis of signaling marker changes within subpopulations to study pathway alterations under different conditions.
• Comparative Analyses: Differential and comparative analyses across diverse biological samples or treatment groups.

Methodology:

Computational methods explicitly include FlowSOM clustering with optional manual cluster merging, dimensionality reduction and heatmaps with dendrograms for cluster reporting, multi-dimensional scaling for quality control, and regression-based differential analyses using GLMM or LMM to model batch effects, paired designs, and overdispersion, implemented in R with Bioconductor packages.