NormExpression

NormExpression evaluates and compares normalization methods for gene expression data to identify approaches that yield consistent results across metrics and dataset types.

Key Features:

• Principle-Based Evaluation: Implements two core principles—consistency of metrics (a method superior by one metric should be superior by another) and consistency across datasets (performance on scRNA-seq should generalize to bulk RNA-seq or microarray).
• Innovative Evaluation Metrics: Uses Area Under normalized CV threshold Curve (AUCVC) to assess performance across varying coefficient of variation thresholds and mSCC (modified Spearman's correlation coefficient) in conjunction with AUCVC for robust method comparison.
• Comprehensive Method Evaluation: Evaluates 14 commonly used normalization methods using scRNA-seq and bulk RNA-seq (and considers microarray) data to test adherence to metric and dataset consistency.

Scientific Applications:

• Normalization method selection: Guides selection of normalization methods to improve reliability of downstream gene expression analyses.
• Single-cell genomics: Provides evaluation metrics tailored to challenges in scRNA-seq normalization.
• Comparative transcriptomics: Assesses method transferability between scRNA-seq, bulk RNA-seq, and microarray datasets for comparative studies.

Methodology:

Applies AUCVC and mSCC to evaluate normalization methods, evaluates 14 normalization methods on scRNA-seq and bulk RNA-seq (and microarray) datasets, and assesses consistency across metrics and datasets.