Systematic Feature Filtering in Exploratory Metabolomics: Application toward Biomarker Discovery

Area of Science:

• Metabolomics
• Analytical Chemistry
• Biomarker Discovery

Background:

• Mass spectrometry-based metabolomics generates numerous features, but over 85% are false positives, hindering biological and clinical interpretation.
• Inefficient elimination of chimeric signals and chemical noise complicates data processing and limits the translational potential of metabolomics.
• Current data processing is a significant bottleneck in metabolomics research.

Purpose of the Study:

• To develop and validate a systematic workflow for refining exploratory metabolomics data and reducing false positives.
• To improve the reliability and interpretability of metabolomics data for biological and clinical applications.
• To streamline data processing for enhanced biomarker discovery.

Main Methods:

• Employed XCMS Online for feature detection, blank subtraction, and reproducibility assessment of liquid chromatography-mass spectrometry data.
• Utilized metabolomics databases for annotation and scrutinized physicochemical properties, comparing predicted and experimental reversed-phase liquid chromatography retention times.
• Developed a retention time-based filtering model and validated significant metabolites using MS/MS acquisition.

Main Results:

• The workflow reduced tentatively identified metabolites by 88%, from 6940 to 839 features.
• Retention time analysis effectively filtered false positives originating from chemical background and chimeric signals.
• Validated a subset of metabolites in a common variable immunodeficiency (CVID) case/control study, confirming potential biomarker structures.

Conclusions:

• The developed data processing workflow effectively removes false positives inherent in mass spectrometry-based metabolomics.
• This systematic approach significantly enhances the quality of metabolomics data, facilitating robust statistical analysis and interpretation.
• The workflow demonstrates broad applicability and streamlines the identification of reliable biomarkers for clinical research.