Differential expression analysis in single-cell and spatial RNA-seq without model assumptions

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell and spatial RNA sequencing technologies have advanced significantly.
• Inconsistent gene up(down)regulation findings persist even in high-quality samples.
• Current data analysis methods may rely on flawed assumptions.

Purpose of the Study:

• To develop a more robust and consistent data analysis approach for RNA sequencing.
• To address inconsistencies in gene expression findings from single-cell and spatial transcriptomics.
• To reduce false-positive and false-negative rates in differential gene expression analysis.

Main Methods:

• Proposed a weighted averaging approach for analyzing transcript counts.
• Incorporated measured noise variances into the weighting of transcript counts.
• Utilized weighted statistical tests instead of standard unweighted tests.
• Related the approach to statistics of cluster-randomized experiments.

Main Results:

• The weighted averaging approach reduces both false-positive and false-negative findings.
• The method eliminates the need for data distribution parametrization and rescaling.
• Analysis becomes less complex and produces more consistent differential gene expression estimates.
• Demonstrated improved consistency in gene expression analysis.

Conclusions:

• The proposed weighted averaging method offers a more reliable analysis for RNA sequencing data.
• This approach enhances the accuracy and consistency of differential gene expression studies.
• It provides a less complex and artifact-free alternative to existing methods.