Evaluation of MassFrontier, MetFrag, MS-FINDER, and SIRIUS for Metabolite Annotation Using an Experimental LC-HRMS...

Area of Science:

• Metabolomics
• Computational Biology
• Biochemistry

Background:

• Untargeted metabolomics offers comprehensive biological profiling.
• Accurate metabolite annotation is hindered by incomplete spectral libraries and structural isomerism.
• In silico annotation tools may enhance coverage but their reliability on real-world data without standards is uncertain.

Purpose of the Study:

• To assess the performance and limitations of four in silico structure prediction tools (MassFrontier, MetFrag, MS-FINDER, SIRIUS/CSI:FingerID).
• To evaluate the reliability of these tools for annotating experimental LC-HRMS data in the absence of reference standards.
• To determine if these tools are sufficient for standalone metabolite annotation.

Main Methods:

• Applied four in silico tools to experimental LC-HRMS data from a study differentiating depressive disorder patients from controls.
• Optimized MS/MS data quality using parallel reaction monitoring for improved fragmentation spectra acquisition.
• Evaluated tool performance based on fragmentation interpretation, candidate generation, and ranking.

Main Results:

• All tested tools suggested structure candidates for most features.
• No single tool was sufficient for reliable metabolite annotation.
• Combined tool application improved annotation over library matching, but often prioritized implausible or artifactual candidates.
• Manual expert evaluation was necessary to confirm chemical plausibility and biological relevance, identifying ten plausible metabolites.

Conclusions:

• In silico annotation tools significantly support metabolomics data annotation but are insufficient alone.
• High-quality MS/MS data acquisition is crucial for reliable metabolite identification.
• Complementary tool usage and mandatory expert curation are essential for accurate metabolite identification in complex biological samples.