Abstract
Despite the increasing availability of tandem mass spectrometry (MS/MS) community spectral libraries for untargeted metabolomics over the past decade, the majority of acquired MS/MS spectra remain uninterpreted. To further aid in interpreting unannotated spectra, we created a nearest neighbor suspect spectral library, consisting of 87,916 annotated MS/MS spectra derived from hundreds of millions of MS/MS spectra originating from published untargeted metabolomics experiments. Entries in this library, or "suspects," were derived from unannotated spectra that could be linked in a molecular network to an annotated spectrum. Annotations were propagated to unknowns based on structural relationships to reference molecules using MS/MS-based spectrum alignment. We demonstrate the broad relevance of the nearest neighbor suspect spectral library through representative examples of propagation-based annotation of acylcarnitines, bacterial and plant natural products, and drug metabolism. Our results also highlight how the library can help to better understand an Alzheimer's brain phenotype. The nearest neighbor suspect spectral library is openly available for download or for data analysis through the GNPS platform to help investigators hypothesize candidate structures for unknown MS/MS spectra in untargeted metabolomics data.
Introduction
In the field of untargeted metabolomics, tandem mass spectrometry (MS/MS) is the gold standard for profiling small molecules in biological samples. However, a persistent challenge remains: on average, only about 5% of acquired spectra can be matched to existing reference libraries. The remaining 95% constitutes biological "dark matter"—molecules that are likely related to known compounds but have not been formally cataloged. This gap significantly limits the biological interpretation of metabolomics data. To address this, we developed a repository-scale "suspect" library that propagates identities from known molecules to their unannotated chemical neighbors.
Methodology
We utilized a hierarchical molecular networking strategy at an unprecedented scale, processing 521 million MS/MS spectra from 1,335 public datasets. The process involved three primary stages:
- Individual Networking: Processing datasets separately to identify reproducible, high-quality spectra.
- Global Co-networking: Linking individual networks into a massive global map to identify chemical connections across diverse studies.
- Annotation Propagation: Identifying unannotated spectra that are "nearest neighbors" to known reference molecules. Identities were assigned based on spectral similarity and predictable mass differences (delta masses).
Results and Discussion
The resulting library contains 87,916 unique MS/MS spectra, representing a significant expansion of the community's toolkit. By incorporating these suspect entries, the spectrum match rate across public datasets increased from approximately 5.5% to 9.3%. In certain environmental samples, the interpretation rate tripled.
The Role of Delta Masses
A key innovation of this library is the identification of 1,350 frequent mass offsets (e.g., the addition of a methyl group or the loss of water). These act as a "Rosetta Stone" for automated naming. For example, an unknown spectrum linked to Acetylcarnitine with a mass difference of 42.022 Da can be identified as an "Acetylcarnitine suspect with addition of $C_3H_6$."
Biological Applications
The library's utility was demonstrated across several domains:
- Alzheimer’s Disease: In a study of 514 human brains, the library identified four previously unannotated acylcarnitines associated with Alzheimer’s, providing a clearer picture of metabolic changes in the diseased brain.
- Drug Metabolism: The library successfully identified drug metabolites in breast milk and natural products in cyanobacteria.
- Environmental Monitoring: The tool helped identify polymers and other contaminants in indoor environments.
Conclusion and Future Work
The nearest neighbor suspect spectral library provides a powerful, open-access resource for interpreting the chemical dark matter of the metabolome. While currently providing Level 3 (family-level) annotations, these "suspect" identities offer critical hypotheses for structural elucidation. Future work will focus on integrating orthogonal data, such as collision cross-section (CCS) values, to further refine these annotations and move toward full structural identification.
Summary updated