MEMO: Mass Spectrometry-Based Sample Vectorization to Explore Chemodiverse Datasets.

Area of Science:

• Natural Products Research
• Computational Metabolomics
• Mass Spectrometry

Background:

• Natural products research involves analyzing diverse biological extracts for novel molecules using liquid chromatography-mass spectrometry (LC-MS/MS).
• Comparing samples analyzed across different batches or time periods is difficult due to experimental variations in LC-MS/MS methods.
• Current metabolomics tools struggle with direct comparison of heterogeneous sample sets, limiting large-scale analysis.

Purpose of the Study:

• To introduce MEMO-MS2 Based Sample Vectorization (MEMO), a novel computational method for clustering diverse samples.
• To enable retention time-agnostic comparison of LC-MS/MS profiles from large, heterogeneous sample sets.
• To overcome limitations in current metabolomics methods for inter-batch sample comparison.

Main Methods:

• Developed MEMO, a method utilizing MS2 spectra for sample vectorization and clustering.
• Applied MEMO to large, chemodiverse sample sets.
• Evaluated MEMO's clustering performance against state-of-the-art metrics.

Main Results:

• MEMO effectively clusters large, chemodiverse sample sets in a retention time-agnostic manner.
• Achieved clustering performance comparable to existing methods that rely on fragmentation spectra.
• Demonstrated significantly reduced computational time and eliminated the need for feature alignment.

Conclusions:

• MEMO provides a robust and efficient solution for comparing complex metabolomics data from multiple samples.
• This method facilitates the analysis of vast sample ensembles across different platforms and time periods.
• MEMO enhances the scope of large-scale comparative analysis in natural products research and computational metabolomics.