Abstract
The growth of online mass spectrometry metabolomics resources, including data repositories, spectral library databases, and online analysis platforms has created an environment of online/web accessibility. Here, we introduce the Metabolomics Spectrum Resolver (https://metabolomics-usi.ucsd.edu/), a tool that builds upon these exciting developments to allow for consistent data export (in human and machine-readable forms) and publication-ready visualisations for tandem mass spectrometry spectra. This tool supports the draft Human Proteome Organizations Proteomics Standards Initiative’s USI specification, which has been extended to deal with the metabolomics use cases. To date, this resource already supports data formats from GNPS, MassBank, MS2LDA, MassIVE, MetaboLights, and Metabolomics Workbench and is integrated into several of these resources, providing a valuable open source community contribution (https://github.com/mwang87/MetabolomicsSpectrumResolver).
Introduction
The effective exchange and visualization of tandem mass spectrometry information across a variety of resources is important in communicating and consequently building confidence in both data quality and molecule identification throughout the research and publication process. The inclusion of MS/MS spectra in scientific posters, presentations, and manuscripts in a consistent manner is often challenging and labor intensive due to the variety of formats and resources available. It is also often difficult to balance both the ease of figure generation and the level of customization possible. On one hand, existing software such as the PDV1, IPSA2 and Lorikeet3 proteomics data viewers as well as vendor software are able to draw MS/MS spectra, but are aimed at interactive visualization. Three key features are often missing: vector graphics (to ensure high resolution export), customization of spectral visualization, and high-throughput automated figure generation. On the other end of the spectrum are generic vector graphics editors such as Adobe Illustrator or Inkscape. While generic editors are powerful, they lack MS/MS specific features and require high levels of customization to achieve an image ready for publication.
We present here the Metabolomics Spectrum Resolver (https://metabolomics-usi.ucsd.edu/), a web tool that enables: 1) high-quality vector graphics drawing, 2) specialised mass spectrometry formatting, 3) integration with major metabolomics data repositories, 4) integration with common online analysis tools and 5) a programmatic web interface (API).
Results and Discussion
MS/MS Data Sources - Integration with Community Resources
This web tool integrates several technologies: spectrum_utils4 for highly customizable spectrum processing/drawing, draft universal spectrum identifiers adapted for Metabolomics inspired by the Human Proteome Organization - Proteome Standards Initiative working group Universal Standards Identifier (USI)5 (http://www.psidev.info/usi), and web APIs supported by online metabolomics data services. The current resource is able to resolve the following metabolomics online services (Table 1):
Data repositories: Metabolomics Workbench6, MetaboLights7, MassIVE8
Reference spectral library resources: MassBank9, MoNA (https://mona.fiehnlab.ucdavis.edu/), GNPS10, and MS2LDA.org MOTIFDB11,
online informatics pipelines: GNPS Molecular Networking/Library Search/MASST/FBMN12, and MS2LDA.org.13
Additionally, several resources (MassBank, GNPS, and MS2LDA.org) have already integrated links to Metabolomics Spectrum Resolver to facilitate the creation of images ready for publication directly from web resources at the time of analysis.
Further, the Metabolomics Spectrum Resolver supports the linking back to the original data resources to allow users to explore the original context of the MS/MS spectra. Due to the web-integrated nature of Metabolomics Spectrum Resolver, after MS/MS publication in manuscripts (generally meant for human consumption), it is straightforward to hyperlink back to the Metabolomics Spectrum Resolver, enabling programmatic access to the underlying MS/MS spectra in a machine readable fashion (text or JSON). This enables re-inspection by both humans and automated tools to re-evaluate identifications with newly available computational tools. Finally, with a programmatic web API interface, software engineers are able to integrate high-throughput figure creation in a language-agnostic fashion. Already, users in the community are using this resource to programmatically pull spectral data in a standardized fashion to facilitate figure generation14.
Finally, the Metabolomics Spectrum Resolver uses spectrum identifiers inspired by the Human Proteome Organizations Proteomics Standards Initiative’s USI specification15. The Metabolomics Spectrum Resolver supports USI formats in the formal specification. Further, the USI specifications have been extended (https://github.com/mwang87/MetabolomicsSpectrumResolver/blob/master/README.md#usi-extended-for-metabolomics---formatting-documentation) to ensure compatibility with a wide range of (at this stage still very heterogeneous) metabolomics resources.
Visualization Capabilities
The Metabolomics Spectrum Resolver can be accessed at https://metabolomics-usi.ucsd.edu/ and enables users to plot a single MS/MS spectrum (Figure 1a) and mirror matches between two MS/MS spectra to show their similarity (e.g. between measured spectra and library matches) (Figure 1b). MS/MS spectra drawing can be modified by specifying mass ranges, intensity ranges, customizable peak labeling, figure size, decimal points for mass values, and an optional grid. The resulting drawing can be downloaded as an SVG (vector), PNG (raster), TSV (tab separated peaks), and JSON (machine readable peaks) (Figure 2).
Conclusion
The Metabolomics Spectrum Resolver has been designed to improve the presentation and accessibility of metabolomics data, both within these supported resources and for the community. The broad support of key online metabolomics resources, programmatic API access, and high quality spectrum drawing make the resource highly accessible and of clear practical benefit to the community and we hope that many in the community will find this useful.
Source Code
Source code can be found here with the MIT license: https://github.com/mwang87/MetabolomicsSpectrumResolver
Conflict of Interests
Mingxun Wang is the founder of Ometa Labs LLC, Pieter C. Dorrestein is on the scientific advisory board of Sirenas and Cybele Microbiome
Acknowledgements
We thank Madeleine Ernst, Alan Jarmusch, and Louis-Felix Nothias for their feedback on the usability of the Metabolomics Spectrum Resolver. ELS acknowledges funding support from the Luxembourg National Research Fund (FNR) for project A18/BM/12341006. MassBank Europe (https://massbank.eu/MassBank) is supported by the NORMAN Association (https://www.norman-network.net), de.NBI (FKZ 031L0107), NaToxAq (Marie Sklodowska-Curie grant agreement No. 722493), HBM4EU (European Union H2020 grant agreement No. 733032) and the Helmholtz Centre for Environmental Research - UFZ (https://www.ufz.de/index.php?en=33573)