NormExpression: An R Package to Normalize Gene Expression Data Using Evaluated Methods.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Data normalization is critical for accurate gene expression analysis.
• Existing evaluation metrics for normalization methods yield inconsistent results, especially for single-cell RNA sequencing (scRNA-seq) data.
• There is a need for established principles to guide the evaluation of normalization methods.

Purpose of the Study:

• To propose principles for evaluating gene expression normalization methods based on metric and dataset consistency.
• To introduce a new metric, Area Under normalized CV threshold Curve (AUCVC), for evaluating normalization methods.
• To provide a framework for selecting optimal normalization methods.

Main Methods:

• Proposed consistency principles: metric consistency and dataset consistency.
• Introduced AUCVC metric and utilized mSCC metric.
• Evaluated 14 normalization methods using scRNA-seq and bulk RNA-seq data.

Main Results:

• The proposed principles were satisfied by the AUCVC and mSCC metrics.
• The study identified robust normalization methods applicable across different data types.
• The R package NormExpression was developed to facilitate method selection.

Conclusions:

• Established principles for reliable evaluation of gene expression normalization methods.
• The AUCVC metric and consistency principles aid in selecting appropriate normalization techniques.
• The NormExpression package offers a practical tool for researchers to evaluate and choose normalization methods.