Feature selection followed by a novel residuals-based normalization simplifies and improves single-cell gene...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) analysis requires normalization to mitigate technical biases and prepare data for statistical analysis.
• Standard workflows typically involve feature selection after normalization to identify highly variable genes (HVGs).

Approach:

• Proposes a revised workflow where feature selection precedes normalization, utilizing observed counts.
• Introduces a novel method for feature selection that identifies both highly variable genes (HVGs) and stable genes.
• Presents a new variance stabilization transformation and residuals-based normalization method informed by stable genes.

Key Points:

• Feature selection can be effectively performed on observed counts before normalization.
• Stable genes identified during feature selection can guide the normalization process.
• The proposed workflow demonstrates significant improvements in downstream clustering analyses.

Conclusions:

• The novel workflow, implemented in the R package Piccolo, offers enhanced accuracy for scRNA-seq data analysis.
• This approach effectively reduces systematic biases and improves the identification of biological variation.
• The method shows robust performance on both simulated and biologically validated datasets.