scKWARN: Kernel-weighted-average robust normalization for single-cell RNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) normalization is crucial for mitigating technical biases like sequencing depth and capture efficiency.
• Current normalization methods often rely on specific data distributions or count-depth relationships, potentially leading to over- or under-correction.

Purpose of the Study:

• To introduce scKWARN, a novel Kernel Weighted Average Robust Normalization method for scRNA-seq data.
• To address limitations of existing methods by correcting technical confounders without restrictive assumptions.

Main Methods:

• scKWARN employs a kernel smoother to generate pseudo-expression profiles by integrating information from neighboring cells.
• Normalization factors are determined by comparing pseudo-profiles to references based on bimodality patterns.

Main Results:

• scKWARN effectively removes diverse technical biases in both simulated and real scRNA-seq datasets.
• The method demonstrates superior performance compared to existing normalization techniques.
• scKWARN successfully preserves genuine biological heterogeneity within the data.

Conclusions:

• scKWARN offers a robust and flexible approach to scRNA-seq data normalization.
• The method enhances the reliability of downstream analyses by accurately correcting technical variations.
• scKWARN is available as an open-source tool for the research community.