CFM-ID 3.0: Significantly Improved ESI-MS/MS Prediction and Compound Identification

Area of Science:

• Analytical Chemistry
• Biochemistry
• Computational Chemistry

Background:

• Metabolite identification in untargeted metabolomics relies heavily on tandem mass spectrometry (MS/MS) spectral data.
• Lack of experimentally collected reference spectra hinders accurate metabolite identification.
• Previous versions of Competitive Fragmentation Modeling-ID (CFM-ID) showed limitations in predicting spectra for certain compound classes, like lipids, and in utilizing available metadata for identification.

Purpose of the Study:

• To significantly improve the performance and speed of CFM-ID for metabolite identification.
• To enhance the accuracy of MS/MS spectral prediction and compound identification capabilities.
• To develop a more robust tool for untargeted metabolomics research.

Main Methods:

• Implemented a rule-based fragmentation approach for improved lipid MS/MS spectral prediction.
• Integrated experimental MS/MS spectra and metadata into the spectral matching algorithm.
• Developed novel scoring functions and a chemical classification algorithm for enhanced accuracy and classification of unknown compounds.

Main Results:

• Achieved significant improvements in CFM-ID's speed and accuracy for MS/MS spectral prediction.
• Enhanced compound identification capabilities by incorporating experimental data and metadata.
• Improved accuracy by 21.1% through new scoring functions.
• Developed a chemical classification algorithm that correctly classifies unknown chemicals in over 80% of cases.

Conclusions:

• CFM-ID 3.0 represents a substantial advancement in computational tools for metabolomics.
• The enhanced prediction accuracy and identification capabilities facilitate more reliable metabolite discovery.
• CFM-ID 3.0 is freely available as a web server and its source code is accessible, promoting wider research application.