Hierarchical Bayesian mixture modelling for antigen-specific T-cell subtyping in combinatorially encoded flow...

Area of Science:

• Biomedical research and immunology
• Computational statistics

Background:

• Automated flow cytometry enables high-throughput protein marker measurement on millions of cells.
• Traditional flow cytometry uses single-color markers, limiting cell subtype identification.
• Combinatorial marker assays with multi-color fluorescent tags enhance cell subtype characterization.

Purpose of the Study:

• To develop and apply novel Bayesian mixture modeling approaches for automated flow cytometry data.
• To address the need for customized statistical methods in immune profiling.
• To improve the identification of functionally differentiated cell subtypes.

Main Methods:

• Development of novel Markov chain Monte Carlo (MCMC) methods.
• Implementation of distributed graphics processing unit (GPU) computation for model fitting.
• Application of a general model framework for cellular subtype identification.

Main Results:

• Demonstrated novel MCMC methods exploiting distributed GPU implementation.
• Provided a detailed example using simulated data for model validation.
• Applied the framework to analyze antigen-specific T-cell subtyping in human blood samples using combinatorial assays.

Conclusions:

• The developed Bayesian mixture modeling framework enhances automated flow cytometry analysis.
• GPU-accelerated MCMC methods offer efficient computation for complex biological data.
• This approach advances the characterization of immune responses and cellular heterogeneity.