WaveICA 2.0: a novel batch effect removal method for untargeted metabolomics data without using batch information

Area of Science:

• Analytical Chemistry
• Bioinformatics
• Metabolomics

Background:

• Untargeted metabolomics using liquid chromatography-mass spectrometry (LC-MS) is susceptible to batch effects, which are systematic biases unrelated to biological variation.
• Existing methods like WaveICA require batch labels, limiting their application when batch information is unavailable or data comprises a single batch.

Purpose of the Study:

• To develop an improved method, WaveICA 2.0, for removing batch effects in untargeted metabolomics data without relying on batch labels.
• To provide a practical R package, WaveICA_2.0, for implementing the enhanced batch effect removal method.

Main Methods:

• The WaveICA method was enhanced to create WaveICA 2.0, enabling batch effect correction without prior batch information.
• An R package, WaveICA_2.0, was developed to facilitate the application of this novel method.

Main Results:

• WaveICA 2.0 demonstrated comparable performance to the original WaveICA for multi-batch data, effectively grouping quality control and subject samples and improving data analysis metrics.
• For single-batch metabolomics data, WaveICA 2.0 effectively removed intensity drift, revealed more biological information, and outperformed QC-RLSC and QC-SVRC methods.

Conclusions:

• WaveICA 2.0 is a practical and effective tool for removing batch effects in untargeted metabolomics data, even when batch information is unknown.
• The method enhances the reliability and biological interpretability of metabolomics datasets, broadening the applicability of batch effect correction techniques.