Observation weights unlock bulk RNA-seq tools for zero inflation and single-cell applications

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for understanding cellular heterogeneity.
• Dropout events in scRNA-seq lead to excessive zero counts, complicating differential expression (DE) analysis.
• Existing dedicated scRNA-seq DE methods have shown limited advantages over traditional bulk RNA-seq tools.

Purpose of the Study:

• To address the challenge of zero inflation in scRNA-seq data.
• To improve the power and accuracy of differential expression analysis in scRNA-seq.
• To enable the effective use of established bulk RNA-seq analysis pipelines for scRNA-seq data.

Main Methods:

• Development of a weighting strategy utilizing a zero-inflated negative binomial model.
• Identification of excess zero counts specific to scRNA-seq data.
• Generation of gene- and cell-specific weights to account for dropout events.

Main Results:

• The proposed weighting strategy effectively identifies and quantifies excess zeros.
• Gene- and cell-specific weights allow the adaptation of bulk RNA-seq DE tools for scRNA-seq data.
• The method significantly boosts the performance of DE analysis on zero-inflated scRNA-seq datasets.

Conclusions:

• A novel weighting strategy can overcome limitations posed by dropout events in scRNA-seq.
• This approach enhances the utility of standard bulk RNA-seq differential expression pipelines for single-cell data.
• The method offers a powerful solution for robust DE analysis in the presence of high zero counts.