scShapes: a statistical framework for identifying distribution shapes in single-cell RNA-sequencing data

Area of Science:

• Genomics
• Computational Biology
• Biostatistics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables quantification of cell-to-cell variation by profiling individual cell transcriptomes.
• Analyzing cell-cell variability in scRNA-seq data is crucial for identifying genes with homogeneous versus heterogeneous expression patterns, moving beyond average expression changes.

Purpose of the Study:

• To introduce scShapes, a novel statistical framework for identifying differential distributions in scRNA-seq data.
• To address limitations of mean-centric analyses by accounting for overdispersion and excess zeros inherent in scRNA-seq data.

Main Methods:

• Utilized generalized linear models within the scShapes framework.
• Quantified gene-specific cell-to-cell variability by testing for differences in expression distributions.
• Flexibly adjusted for covariates as needed.

Main Results:

• scShapes identified subtle expression variations independent of altered mean expression.
• The framework detected biologically relevant genes missed by standard differential expression approaches.
• Highlighted genes transitioning from unimodal to zero-inflated distributions, suggesting mechanisms like transcriptional bursting.

Conclusions:

• scShapes expands understanding of gene expression's role in transcriptional regulation and cellular phenotypes.
• The framework provides a powerful tool for analyzing complex single-cell expression patterns.
• scShapes is available as a Bioconductor R package for broader accessibility.