immunoClust

immunoClust performs model-based clustering and meta-clustering of high-dimensional fluorescence and mass cytometry data to identify and classify cell populations, including rare subsets, for biomarker discovery and comparative analyses of up to 120 single-cell parameters.

Key Features:

• Automated Clustering: Model-based clustering using an iterative expectation maximization algorithm for unsupervised grouping of high-dimensional cell events.
• Meta-Clustering for Population Classification: A classification/meta-clustering algorithm that assigns clusters into comparable populations across multiple samples.
• Sensitivity to Rare Cell Types: Capability to detect rare cell populations present alongside large populations.
• Uncompensated Data Analysis: Ability to process uncompensated fluorescence and mass cytometry data without prior compensation steps.
• High-Dimensional Support: Designed to handle datasets with up to 120 parameters per single cell.

Scientific Applications:

• Biomarker Discovery: Enables detailed analysis of peripheral blood cell populations to identify potential biomarkers associated with chronic inflammatory disorders.
• Personalized Medicine: Standardizes cytometric data analysis to support development of individualized immune-profile-based therapeutic recommendations.
• Benchmarking and Method Validation: Used for benchmarking and validation with blood cell samples of known composition, FlowCAP III challenge datasets, and high-dimensional fluorescence and mass-cytometry datasets compared with manual gating.

Methodology:

Global unsupervised model-based clustering of cell events using an iterative expectation maximization algorithm leveraging integrated classification likelihood, followed by a classification/meta-clustering algorithm that assigns clusters into comparable populations across samples.