TWO-SIGMA-G: a new competitive gene set testing framework for scRNA-seq data accounting for inter-gene and cell-cell...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) generates complex, high-dimensional data.
• Accurate gene set testing is crucial for interpreting scRNA-seq results.
• Existing methods may struggle with complex experimental designs and inter-gene correlations.

Purpose of the Study:

• To introduce TWO-SIGMA-G, a novel competitive gene set test for scRNA-seq data.
• To improve statistical rigor and power in gene set analysis.
• To control the false positive rate at the set level, especially with inter-gene correlation.

Main Methods:

• Utilizes a mixed-effects regression model, extending the TWO-SIGMA framework.
• Addresses complex experimental designs and correlations between biological replicates.
• Incorporates a novel adjustment for inter-gene correlation (IGC) at the set level.

Main Results:

• Simulations show TWO-SIGMA-G preserves type-I error rates.
• Demonstrates increased statistical power in the presence of IGC compared to other methods.
• Successfully identified HIV-associated interferon pathways and Alzheimer's disease pathways in real datasets.

Conclusions:

• TWO-SIGMA-G offers a flexible and rigorous approach for gene set testing in scRNA-seq.
• The method effectively handles complex data structures and controls for IGC.
• Provides valuable insights into biological pathways in both animal models and human diseases.