BayesRare: Bayesian mixture model for population-level rare cell type detection in multi-subject single-cell RNA...

Area of Science:

• Genomics
• Computational Biology
• Biostatistics

Background:

• Rare cell types in single-cell RNA sequencing (scRNA-seq) data are crucial for understanding biological signals like disease markers and immune regulation.
• Large-scale scRNA-seq cohorts allow population-level analysis of rare cells, but existing detection methods struggle with cross-subject data integration.

Purpose of the Study:

• To develop a novel framework, BayesRare, for discovering rare cell populations across multiple subjects in scRNA-seq data.
• To enable population-level analysis of rare cell prevalence, heterogeneity, and disease associations by leveraging cross-subject information.

Main Methods:

• BayesRare employs a hierarchical Bayesian framework with a mixture model and a rare cluster indicator for joint cell-type clustering and rare-population identification.
• The method integrates evidence across subjects, quantifies uncertainty using posterior probabilities, and enables group-level difference inference.

Main Results:

• BayesRare demonstrated superior precision and reduced false positives compared to existing methods on synthetic and real scRNA-seq datasets.
• The framework successfully uncovered biologically meaningful, disease-specific rare cell subtypes.

Conclusions:

• BayesRare provides a robust statistical approach for population-level rare cell discovery in multi-subject scRNA-seq data.
• This method enhances the ability to identify and characterize rare cell populations, advancing our understanding of their role in disease.