A Guideline to Univariate Statistical Analysis for LC/MS-Based Untargeted Metabolomics-Derived Data

Area of Science:

• Metabolomics
• Analytical Chemistry
• Bioinformatics

Background:

• LC/MS-based metabolomics relies on software for feature detection and quantification.
• Statistical analysis is crucial for identifying significantly altered metabolite features between samples.
• Unambiguous metabolite identification requires comparing retention time and MS/MS data with standards.

Purpose of the Study:

• To provide a comprehensive overview of a statistical analysis workflow for ranking metabolite features.
• To guide the selection of features for subsequent MS/MS experiments.
• To discuss the characteristics and challenges of univariate data analysis in LC/MS metabolomics.

Main Methods:

• Focus on univariate data analysis applied to all detected features.
• Utilized four different real LC/MS untargeted metabolomic datasets for illustration.
• Demonstrated the impact of adhering to or violating statistical assumptions of univariate tests.

Main Results:

• Illustrated the influence of mathematical assumptions (normality, homocedasticity) on statistical test outcomes.
• Discussed critical data analysis issues including sample size, analytical variation, and multiple testing correction.
• Provided practical insights using real-world LC/MS datasets.

Conclusions:

• A robust statistical workflow is essential for reliable metabolite feature ranking in LC/MS metabolomics.
• Understanding and addressing statistical assumptions is critical for accurate data interpretation.
• The presented methods aid in selecting relevant features for definitive metabolite identification.