CoChat
Sign Up
PreprintCited by 79Glycosylation and Glycoproteins ResearchMass Spectrometry Techniques and ApplicationsAdvanced Proteomics Techniques and Applications

The DOI matches its Crossref record; 1 of the 1 references it cites were found in a registry or an index.

Predicting glycan structure from tandem mass spectrometry via deep learning

Open access
Urban, James, Jin, Chunsheng, Thomsson, Kristina A., Karlsson, Niclas G., Ives, Callum M., Fadda, Elisa, Bojar, Daniel·2023·Nature Methods
DOI 10.1101/2023.06.13.544793PMC PMC11239490

Preprint — published as 10.1038/s41592-024-02314-6 · Paper page

Open this paper in CoChat — free accountOpen PDF

Abstract

Glycans constitute the most complicated post-translational modification, modulating protein activity in health and disease. However, structural annotation from tandem mass spectrometry data is a bottleneck in glycomics, preventing high-throughput endeavors and relegating glycomics to a few experts. Trained on a newly curated set of 300,000 annotated MS/MS spectra, we present CandyCrunch, a dilated residual neural network predicting glycan structure from raw LC-MS/MS data in seconds (Top1 Accuracy: 87.7%). We developed an open-access Python-based workflow of raw data conversion and prediction, followed by automated curation and fragment annotation, with predictions recapitulating and extending expert annotation. We demonstrate that this can be used for de novo annotation, diagnostic fragment identification, and high-throughput glycomics. For maximum impact, this entire pipeline is tightly interlaced with our glycowork platform and can be easily tested at https://colab.research.google.com/github/BojarLab/CandyCrunch/blob/main/CandyCru nch.ipynb . We envision CandyCrunch to democratize structural glycomics and the elucidation of biological roles of glycans.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1101/2023.06.13.544793

Cites

1 of 1 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 55

  1. 2022, Cancer glycomics offers potential biomarkers and therapeutic targets in the framework of 3P medicine, Front. Endocrinol, 13, 970489, 10.3389/fendo.2022.970489 · DOI
  2. 10.1136/bmjdrc-2021-002263 · DOI
  3. 10.1093/glycob/cww086 · DOI · Paper page
  4. 2022, Influence of Host Sialic Acid Receptors Structure on the Host Specificity of Influenza Viruses, Viruses, 14, 2141, 10.3390/v14102141 · DOI
  5. Rudd, P. M. et al. Glycomics and Glycoproteomics. in Essentials of Glycobiology (eds. Varki, A. et al.) (Cold Spring Harbor Laboratory Press, 2022).
  6. 2021, Correcting for sparsity and interdependence in glycomics by accounting for glycan biosynthesis, Nat Commun, 12, 4988, 10.1038/s41467-021-25183-5 · DOI
  7. 10.1101/2023.02.04.527106 · DOI
  8. 10.1007/s13361-013-0610-4 · DOI
  9. 2017, Characterization of isomeric glycan structures by LC-MS/MS: Liquid Phase Separations, ELECTROPHORESIS, 38, 2100, 10.1002/elps.201700042 · DOI
  10. 2022, High-Throughput Glycomic Methods, Chem. Rev, 122, 15865, 10.1021/acs.chemrev.1c01031 · DOI
  11. 10.1002/mas.21806 · DOI
  12. 10.1021/ac0010126 · DOI
  13. 2016, Distinctive MS/MS Fragmentation Pathways of Glycopeptide-Generated Oxonium Ions Provide Evidence of the Glycan Structure, Chem. Eur. J, 22, 1114, 10.1002/chem.201503659 · DOI
  14. 2019, Cross-Ring Fragmentation Patterns in the Tandem Mass Spectra of Underivatized Sialylated Oligosaccharides and Their Special Suitability for Spectrum Library Searching, J. Am. Soc. Mass Spectrom, 30, 426, 10.1007/s13361-018-2106-8 · DOI
  15. 2022, HexNAcQuest: A Tool to Distinguish O-GlcNAc and O-GalNAc, J. Am. Soc. Mass Spectrom, 33, 2008, 10.1021/jasms.2c00172 · DOI
  16. 2016, Classification of Tandem Mass Spectra for Identification of N- and O-linked Glycopeptides, Sci Rep, 6, 37189, 10.1038/srep37189 · DOI
  17. 2017, Glycoforest 1.0, Anal. Chem, 89, 10932, 10.1021/acs.analchem.7b02754 · DOI
  18. 2022, GlycoDeNovo2: An Improved MS/MS-Based De Novo Glycan Topology Reconstruction Algorithm, J. Am. Soc. Mass Spectrom, 33, 436, 10.1021/jasms.1c00288 · DOI
  19. 2015, A Machine Learning Based Approach to de novo Sequencing of Glycans from Tandem Mass Spectrometry Spectrum, IEEE/ACM Trans. Comput. Biol. and Bioinf, 12, 1267, 10.1109/TCBB.2015.2430317 · DOI
  20. 10.1021/pr7008252 · DOI
  21. 10.1021/acs.jproteome.9b00367 · DOI
  22. 2009, Software Platform for High-Throughput Glycomics, Anal. Chem, 81, 3252, 10.1021/ac802408f · DOI
  23. 10.1101/2023.01.03.522621 · DOI
  24. 2022, Ad hoc learning of peptide fragmentation from mass spectra enables an interpretable detection of phosphorylated and cross- linked peptides, Nat Mach Intell, 4, 378, 10.1038/s42256-022-00467-7 · DOI
  25. 2022, Computational Modeling of O-Linked Glycan Biosynthesis in CHO Cells, Molecules, 27, 1766, 10.3390/molecules27061766 · DOI
  26. 10.1101/2023.05.20.541563 · DOI
  27. 10.1016/j.chom.2020.10.004 · DOI
  28. 2017, Structural Diversity of Human Gastric Mucin Glycans, Molecular & Cellular Proteomics, 16, 743, 10.1074/mcp.M117.067983 · DOI
  29. 10.1101/2023.01.16.524336 · DOI
  30. 10.1093/glycob/cwab067 · DOI
  31. 10.1007/BF01049915 · DOI
  32. 10.1093/bioinformatics/btr137 · DOI
  33. Guo, C. , Pleiss, G. , Sun, Y. & Weinberger, K. Q. On Calibration of Modern Neural Networks. Preprint at http://arxiv.org/abs/1706.04599 (2017). · arXiv
  34. 10.1016/j.jasms.2006.02.022 · DOI
  35. 2022, Profiling the glycome of Cardicola forsteri, a blood fluke parasitic to bluefin tuna, International Journal for Parasitology, 52, 1, 10.1016/j.ijpara.2021.06.004 · DOI
  36. 2021, Community evaluation of glycoproteomics informatics solutions reveals high-performance search strategies for serum glycopeptide analysis, Nat Methods, 18, 1304, 10.1038/s41592-021-01309-x · DOI
  37. 1999, OPTICS: ordering points to identify the clustering structure, SIGMOD Rec, 28, 49, 10.1145/304181.304187 · DOI
  38. 2021, Integrated N- and O-Glycomics of Acute Myeloid Leukemia (AML) Cell Lines, Cells, 10, 3058, 10.3390/cells10113058 · DOI
  39. 2023, O-Glycomic and Proteomic Signatures of Spontaneous and Butyrate-Stimulated Colorectal Cancer Cell Line Differentiation, Molecular & Cellular Proteomics, 22, 100501, 10.1016/j.mcpro.2023.100501 · DOI
  40. 10.1093/bioinformatics/btu308 · DOI
  41. 2014, APP: an Automated Proteomics Pipeline for the analysis of mass spectrometry data based on multiple open access tools, BMC Bioinformatics, 15, 441, 10.1186/s12859-014-0441-8 · DOI
  42. 10.1093/nar/gkaa1012 · DOI
  43. 10.1038/nprot.2007.422 · DOI
  44. 10.48550/ARXIV.2301.11419 · DOI
  45. 2017, Proteomics, 1550, 339, 10.1007/978-1-4939-6747-6_23 · DOI
  46. 10.1093/bioinformatics/bty046 · DOI
  47. Paszke, A. et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library. arXiv:1912.01703 [cs, stat] (2019). · arXiv
  48. 10.48550/ARXIV.2102.11600 · DOI
  49. 2021, MS2DeepScore: a novel deep learning similarity measure to compare tandem mass spectra, J Cheminform, 13, 84, 10.1186/s13321-021-00558-4 · DOI
  50. 10.48550/ARXIV.2204.12511 · DOI
  51. 2005, A Faster Algorithm for Detecting Network Motifs. in Algorithms in Bioinformatics, 3692, 165
  52. 10.1002/jcc.20820 · DOI
  53. 10.1002/jcc.20035 · DOI
  54. 10.1186/1756-0500-5-367 · DOI
  55. Abraham, M. J. et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25 (2015). · DOI · Paper page

Metadata

Authors
Urban, J., Jin, C., Thomsson, K. A., Karlsson, N. G., Ives, C. M., Fadda, E., & Bojar, D.
Affiliations
  • Wallenberg Centre for Molecular and Translational Medicine (SE)
  • University of Gothenburg (SE)
  • OsloMet – Oslo Metropolitan University (NO)
  • National University of Ireland, Maynooth (IE)
  • University of Southampton (GB)
Corresponding author
Bojar, Daniel
ORCID
Venue
Nature Methods
Publisher
openRxiv
Published
2023-06-13
Volume
21
Issue
7
Pages
1206-1215
ISSN
1548-7091 (ISSN-L)
DOI
10.1101/2023.06.13.544793
PMC
PMC11239490
OpenAlex
W4400208210
Cited by
79
References
1
Citations by year
2026: 36 · 2025: 26 · 2024: 14 · 2023: 3
FWCI
10.35
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Nature Methods (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nd)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • ePrints Soton (University of Southampton) (acceptedVersion, other-oa)
Funders
  • Science Foundation Ireland (20/FFP-P/, 20/FFP-P/8809)
  • Knut och Alice Wallenbergs Stiftelse
  • Vetenskapsrådet
  • Göteborgs Universitet
  • Science for Life Laboratory
  • Sahlgrenska Akademin
  • Fogarty International Center
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Glycosylation and Glycoproteins Research
  • Physical Sciences › Chemistry › Spectroscopy › Mass Spectrometry Techniques and Applications
  • Physical Sciences › Chemistry › Spectroscopy › Advanced Proteomics Techniques and Applications
Keywords
glycan structure prediction, tandem mass spectrometry, deep learning, glycomics, fragment identification, post-translational modification
MeSH terms
Deep Learning, Chromatography, Liquid, Humans, Polysaccharides, Software, Neural Networks, Computer, Tandem Mass Spectrometry, Glycomics, Workflow

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review