Back

Large-scale discovery and annotation of hidden substructure patterns in mass spectrometry profiles

Torres Ortega, L. R.; Dietrich, J.; Wandy, J.; Mol, H.; van der Hooft, J. J. J.

2025-07-11 biochemistry
10.1101/2025.06.19.659491 bioRxiv
Show abstract

Untargeted mass spectrometry measures many spectra of unknown molecules. To annotate them, tandem mass spectrometry (MS/MS) generates fragmentation patterns representing common substructures. MS2LDA discovers these patterns via unsupervised topic modelling as Mass2Motifs. However, MS/MS-based substructure identification is limited by computational efficiency and interpretability. Here, we report an up to 14x speed improvement through improved algorithmic efficiency. Furthermore, the new automated Mass2Motif Annotation Guidance (MAG) aids in structurally identifying Mass2Motifs. Using three chemically diverse curated MotifDB-MotifSets for benchmarking, MAG achieved median substructure overlap scores of 0.75 up to 0.93, demonstrating robust substructural annotations. We further validated MS2LDA 2.0 in experimental data by identifying substructures of pesticides spiked into a biological matrix and demonstrated its discovery potential by annotating previously uncharacterized fungal natural products. Together with the new visualization app and MassQL-searchable MotifDB, we anticipate that MS2LDA 2.0 will boost the identification of novel chemistry and hidden patterns in mass spectrometry profiles.

Published in Nature Communications (predicted rank #2) · training set

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.