MAST

MAST models single-cell RNA sequencing (scRNA-seq) data using a two-part generalized linear model to account for zero-inflation and bimodal gene expression, facilitating gene set enrichment and co-expression network analyses.

Key Features:

• Two-Part Generalized Linear Model (GLM): Employs a two-part GLM that parameterizes the bimodal distribution of expression to model zero-inflation and non-zero expression levels.
• Adjustment for Cellular Detection Rate: Adjusts for cellular detection rate, defined as the fraction of genes expressed in a cell, as a nuisance covariate to reduce technical variation.
• Gene Set Enrichment Analysis (GSEA): Implements gene set enrichment analysis tailored for single-cell data to identify significant biological pathways and processes within heterogeneous cell populations.
• Gene Network Evolution Insights: Facilitates exploration of how networks of co-expressed genes evolve across different experimental treatments to reveal dynamic regulatory responses.

Scientific Applications:

• Single-cell transcriptomics analysis: Analyzing scRNA-seq datasets to characterize gene expression heterogeneity and bimodal expression patterns.
• Development and differentiation studies: Investigating developmental processes and cellular differentiation by resolving cell-type-specific expression changes.
• Disease progression profiling: Studying disease progression by profiling cellular responses and expression dynamics across conditions.
• Pathway and network characterization for precision medicine: Identifying pathway-level and co-expression network changes relevant to precision medicine and personalized therapeutic strategies.

Methodology:

Computational methods include a two-part generalized linear model, inclusion of cellular detection rate as a covariate, single-cell–tailored gene set enrichment analysis, and analysis of co-expression network changes across treatments.