Tailor

Tailor performs model-based clustering of high-parameter flow cytometry data using a phenotype-aware binning scheme and multivariate Gaussian mixture modeling to identify and annotate cellular populations for comparative immunophenotyping and longitudinal studies.

Key Features:

• Phenotype-aware binning scheme: Implements an initial coarse, phenotype-sensitive binning step to manage complexity in datasets containing tens of millions of flow cytometry events.
• Multivariate Gaussian mixture modeling: Refines the coarse model with multivariate Gaussian mixture models optimized to handle heavy-tailed distributions and inter-sample variability.
• Robustness and benchmarking: Demonstrates resilience to moderate departures from Gaussian assumptions and sample variation as assessed by simulation studies and real-world datasets.
• Automated non-overlapping annotations: Produces automatic, non-overlapping cluster annotations intended to be interpretable for downstream analysis.

Scientific Applications:

• Comparative immunophenotyping: Enables consistent identification and comparison of cellular populations across different samples or experimental conditions.
• Longitudinal studies: Supports tracking of cellular population changes over time by being robust to inter-sample variation.
• High-dimensional data analysis: Facilitates clustering and interpretation of complex, high-parameter flow cytometry datasets to investigate cellular heterogeneity and function.

Methodology:

Integrates a phenotype-aware binning scheme to create an initial coarse model that is refined by multivariate Gaussian mixture modeling, explicitly accounting for heavy-tailed distributions in flow cytometry data.