Consensus ranking and fragmentation prediction for identification of unknowns in high resolution mass spectrometry
0 likes227 views
This document summarizes a presentation about using the EPA's CompTox Chemistry Dashboard to improve the identification of unknown chemicals detected by high resolution mass spectrometry (HRMS). The Dashboard integrates data from multiple sources to rank candidate identifications based on factors like reference counts. It also allows prediction and matching of MS/MS fragmentation patterns to experimentally observed data, improving identification confidence. Future work includes combined data visualization, predicted retention times, and ongoing database expansion to further enhance the Dashboard's utility for non-target analysis workflows.
Consensus ranking and fragmentation prediction for identification of unknowns in high resolution mass spectrometry
- 1. Photo image area measures 2” H x 6.93” W and can be masked by a collage strip of one, two or three images. The photo image area is located 3.19” from left and 3.81” from top of page. Each image used in collage should be reduced or cropped to a maximum of 2” high, stroked with a 1.5 pt white frame and positioned edge-to-edge with accompanying images. Consensus ranking and fragmentation prediction for identification of unknowns in high resolution mass spectrometry Office of Research and Development National Center for Computational Toxicology, RTP, NC August 20, 2018 Andrew D. McEachran Hussein Al-Ghoul, Ilya Balabin, Tommy Cathey, Alex Chao, Jon Sobus, and Antony J. Williams http://orcid.org/0000-0003-1423-330X The views expressed in this presentation are those of the author and do not necessarily reflect the views or policies of the U.S. EPA AGRO 107
- 2. General Goals of SSA/NTA - 1 Dust Sample - Negative Ionization Mode - 300 Extracted “Molecular Features” 1) Prioritize “Molecular Features” 2) Correctly assign formulas 3) Correctly assign structures 4) Identify chemical sources 5) Predict chemical concentrations C17H19NO3 12 µg/g (1) (2) (3) (4) (5) EXPOSURE 8/20/18
- 4. C17H19NO3…. 8/20/18 Schymanski, et al. 2014 plus 100s more…
- 5. 8/20/18 Analytical Instruments Comp. Tools & Workflows Databases The General Approach
- 6. 8/20/18 Analytical Instruments Comp. Tools & Workflows Databases The General Approach
- 8. 8/20/18
- 9. 8/20/18
- 10. 8/20/18
- 11. 8/20/18
- 12. 8/20/18 C10H14N2 Batch Search for SSA/NTA
- 13. MS-Ready Structures improve database searching 8/20/18 McEachran, et al, accepted manuscript
- 14. MS-Ready Structures improve database searching 8/20/18 McEachran, et al, accepted manuscript
- 16. Data Source Ranking of “known unknowns” • Mass and/or formula unknown to a researcher, contained within a reference database • Most likely candidate chemicals have the most references/sources 8/20/18 C14H22N2O3 266.16304 Chemical Reference Database Sorted candidate structures
- 17. Initial Data Source Ranking in ChemSpider 8/20/18 • Adopted by NTA researchers around the world
- 18. • On same 162 chemicals, Dashboard outperforms ChemSpider 8/20/18
- 19. Additional Data Streams to Improve Identifications • US EPA CompTox Dashboard Data Sources (DS) • PubChem Data Source Count • PubMed Reference Count • Presence in STOFF-IDENT Database • Predicted Environmental Media Occurrence • OPERA PhysChem Properties • NORMAN Network Priority List 8/20/18
- 20. 8/20/18 All available via Batch Search:
- 21. 8/20/18 Identification ranks for 1783 chemicals using multiple data streams 𝑆𝑆𝑆𝑆𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 = 𝑆𝑆𝑆𝑆𝐷𝐷𝐷𝐷 + 𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 + 𝑆𝑆𝑆𝑆𝑅𝑅𝑅𝑅 + 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀 + ⋯
- 22. Metadata is critical, but need structural confirmation to increase confidence 8/20/18 Schymanski, et al. 2014
- 23. Library Fragmentation Spectra (20eV) Observed Fragmentation Spectra (20eV) Match Score MS/MS Spectral Matching for Identification 8/20/18
- 24. CFM-ID • Fragmentation prediction for identification in HRMS • Open source code allows for MS/MS spectra prediction for ESI+, ESI-, and EI • Predictions generated and stored for >700,000 structures, to be accessible via CompTox Dashboard • Python code to pull matches and score experimental vs predicted spectra • Cosine dot product match score calculation 8/20/18 Allen, et al 2014, 2015, 2016
- 25. 8/20/18
- 26. 8/20/18 Predicted MS/MS spectra provide greater coverage than empirical libraries Chao, et al, in prep
- 27. Evaluating on CASMI 2016 • Critical Assessment of Small Molecule Identification –Training data= 312 peak lists (from 285 substances) • 234 MS/MS in positive mode • 58 in negative mode –Challenge Data= 208 peak lists (from 188 substances) • 127 in positive mode • 81 in negative mode • Precursor ion search window= 15 ppm • Fragment ion match threshold= 0.02 Da • Candidates limited to Dashboard results within precursor ion search window 8/20/18 http://www.casmi-contest.org/2016/index.shtml
- 28. 8/20/18 # Identified % of Total #1 Hits 89 43% Top 5 154 74% Top 10 174 84% Top 20 190 91% # Identified % of Total #1 Hits 154 74% Top 5 195 94% Top 10 198 95% Top 20 202 97% CFM-ID only CFM-ID +DSSTox Data Sources CASMI 2016 Contest Challenge Set (n=208)
- 29. Access via CompTox Dashboard 8/20/18 • Data available for download after publication McEachran, et al, in prep
- 30. 8/20/18 Mockup, work in progress….
- 31. Dashboard in NTA Workflows Identification by Data Source Ranking Retention Time Prediction MS/MS Data and Fragmentation Prediction Environmental Media Occurrence Functional Use/Product Occurrence 8/20/18 CompTox Dashboard
- 32. Future Directions • Combined data visualization • Retention time index (RTI) predictions • Ongoing expansion of the database • Integration to public MS databases 8/20/18 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 A C N M e th o d E xp e rim e n ta l R T I PredictedRTI
- 33. Conclusions • Databases are effective resources in SSA/NTA • CompTox Dashboard provides access to chemistry data for >760,000 chemical substances • Predicted MS/MS spectra linked within the CompTox Dashboard further enhances effectiveness and increases confidence in identifications 8/20/18
- 34. Acknowledgements EPA NCCT Tony Williams Chris Grulke Jeff Edwards EPA NERL Katherine Phillips Kristin Isaacs Kathie Dionisio Jon Sobus Mark Strynar Elin Ulrich Seth Newton 8/20/18 External Collaborators Emma Schymanski- Univ. Luxembourg Christoph Ruttkies- IPB, Halle Kamel Mansouri- ILS, Inc
- 35. Questions? • mceachran.andrew@epa.gov • http://orcid.org/0000-0003-1423-330X • Associated presentations: –AGRO 29: Leveraging chemistry data to improve exposure analyses using the EPA’s CompTox Chemistry Dashboard –ANYL 100: Developing tools for high resolution mass spectrometry-based screening via the EPA’s CompTox Chemistry Dashboard –ENVR 152: EPA Comptox Chemistry Dashboard as a data integration hub for environmental chemistry data 8/20/18