DecoID improves identification rates in metabolomics through database-assisted MS/MS deconvolution.

Area of Science:

• Metabolomics
• Computational Chemistry
• Mass Spectrometry

Background:

• Chimeric MS/MS spectra, resulting from multiple precursor ions, pose a significant challenge for accurate compound identification in metabolomics.
• Existing methods for deconvoluting chimeric spectra often rely on specific experimental techniques, introducing variability and limiting their applicability.

Purpose of the Study:

• To develop a novel, method-independent computational approach for deconvoluting chimeric MS/MS spectra.
• To enhance metabolite identification rates in metabolomics datasets acquired through various mass spectrometry methods.

Main Methods:

• DecoID utilizes LASSO regression to computationally mix database spectra, enabling a match to experimentally acquired chimeric spectra.
• The approach is designed to be independent of specific experimental acquisition methods, offering broad compatibility.

Main Results:

• DecoID significantly increases the number of identified metabolites in both data-independent and data-dependent acquisition MS/MS datasets.
• The method demonstrated a >30% increase in metabolite identification from human plasma data compared to traditional spectral matching.
• No increase in the false discovery rate was observed, ensuring the reliability of the identified metabolites.

Conclusions:

• DecoID offers a robust and versatile solution for deconvoluting chimeric MS/MS spectra, overcoming limitations of previous methods.
• The tool enhances the depth of metabolite profiling in metabolomics studies and is compatible with diverse MS/MS databases.