Unknown Metabolite Identification Using Machine Learning Collision Cross-Section Prediction and Tandem Mass...

Area of Science:

• Metabolomics
• Analytical Chemistry
• Biochemistry

Background:

• Ion mobility (IM) spectrometry offers complementary data to mass spectrometry (MS) for metabolite identification.
• Current IM-MS applications often focus on ideal conditions, with limited evaluation in complex biological matrices.

Purpose of the Study:

• To develop and evaluate a workflow for identifying unknown differential metabolites in complex samples using IM-MS.
• To assess the utility of de novo molecular formula annotation and MS/MS structure elucidation with experimental and predicted collision cross-section (CCS) data.

Main Methods:

• Utilized SIRIUS 4 for molecular formula annotation and MS/MS structure elucidation.
• Integrated experimental IM CCS measurements and machine learning CCS predictions.
• Applied the workflow to analyze differential metabolites in mutant Caenorhabditis elegans strains.

Main Results:

• The workflow successfully filtered candidate structures for some ion features.
• Instrumentation performance and data analysis presented significant challenges.
• Only 37% of differential features yielded both MS/MS and CCS data.
• CCS filtering reduced candidate structures by an average of 28% with a ±3% error cutoff.
• Limitations included poor machine learning training set matching and inaccurate CCS values.

Conclusions:

• IM-MS shows promise for unknown metabolite identification in non-targeted metabolomics.
• Significant bottlenecks exist in instrumentation, data analysis, and CCS prediction accuracy.
• Further improvements are needed to enhance the reliability and efficiency of IM-MS for complex sample analysis.