TRIANGLE AREA MASS SPECTOMETRY MEETING: Structure Identification Approaches Using the US-EPA CompTox Chemicals Dashboard to Support Mass Spectrometry Analyses

TRIANGLE AREA MASS SPECTOMETRY MEETING: Structure Identification Approaches Using the US-EPA CompTox Chemicals Dashboard to Support Mass Spectrometry Analyses

1. 1. Structure Identification Approaches Using the US-EPA CompTox Chemicals Dashboard to Support Mass Spectrometry Analyses Triangle Area Mass Spectrometry RTP, January 2020 http://www.orcid.org/0000-0002-2668-4821 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 Antony Williams Center for Computational Toxicology and Exposure, US-EPA, RTP, NC …and an enormous cast of characters
2. 2. Outline • Quick overview of the dashboard • Specific data of interest to this audience (it’s not just Computational Toxicology) • Support for Mass Spectrometry • Data quality in the public domain • Work in progress – prototypes • A request for help 1
3. 3. 2 SEARCH TOX DATA BIOACTIVITY SIMILARITY READ-ACROSS PUBMED BATCH SEARCH CompTox Chemicals Dashboard https://comptox.epa.gov/dashboard
4. 4. BASIC Search 3
5. 5. Detailed Chemical Pages 4
6. 6. Properties, Fate and Transport 5
7. 7. Properties, Fate and Transport e.g. Solubility 6
8. 8. Properties, Fate and Transport e.g. logP 7
9. 9. Sources of Exposure to Chemicals 8
10. 10. Identifiers to Support Searches 9
11. 11. Link Access 10
12. 12. Mass Spec Links 11
13. 13. NIST WebBook https://webbook.nist.gov/chemistry/ 12
14. 14. MassBank of North America https://mona.fiehnlab.ucdavis.edu 13
15. 15. Batch Searching 14
16. 16. Aggregate data for a list of chemicals 15
17. 17. Batch Search Names 16 Excel Download
18. 18. Add Other Data of Interest 17
19. 19. Chemical Lists of Interest… 18
20. 20. 225 Chemical Lists (and growing) 19
21. 21. “Volatilome” Human Breath 20
22. 22. “Volatilome” Saliva 21
23. 23. PFAS lists of Chemicals 22
24. 24. Building a “reference” PFAS list • PFAS structure list (PFASSTRUCT) is expanded from public databases, EPA agency lists and literature • Approaching ~7000 structures – 98.8% have associated CAS Numbers • Compare with PubChem 220,720 structures 23
25. 25. Formula Search can find isomers 24
26. 26. Active expansion of the PFAS list From 2 to 8 variants of PFOS 25
27. 27. Disinfection By-Products 26
28. 28. Mycotoxins • Two lists: 328 and 88 members 27
29. 29. Vomitoxin 28
30. 30. BIG databases are GREAT! P u b C h e m C A S R e g is try C h e m S p id e r E P A D S S T o x B lo o d E x p o s o m e 1 0 4 1 0 5 1 0 6 1 0 7 1 0 8 1 0 9 ChemicalSubstances • Thanks to all of the public database efforts • So much benefit from what’s been done • There are hundreds of them at this point…
31. 31. Vomitoxin - ChemSpider • 19 “Vomitoxins” – 3 isotopically labeled 30
32. 32. Vomitoxin – PubChem 31 • 33 unique InChI Keys
33. 33. PubChem – “virtual chemistry” • Other databases grow quickly…a lot of “virtual chemistry” and “make on demand” compounds. Vomitoxin has 7 ZINC stereoforms. • The Dashboard database grows slowly (next release is +20k chemicals in 6 months) 32
34. 34. ChemSpider – lots of virtuals??? 33 • 52 million chemicals from one vendor
35. 35. Taxol: 79 Results 34
36. 36. Data Quality is important • Data quality in free web-based databases! 35
37. 37. We’re still cleaning data too 36
38. 38. Tire Crumb Rubber (298) 37
39. 39. Terpenes in Vape (37) 38
40. 40. Hydraulic Fracturing (1640) 39
41. 41. Opioids and Metabolites (160) 40
42. 42. “MS-ready” structures 41
43. 43. Overview of MS-Ready Structures • All structure-based chemical substances are algorithmically processed to – Split multicomponent chemicals into individual structures – Desalt and neutralize individual structures – Remove stereochemical bonds from all chemicals • MS-Ready structures are then mapped to original substances to provide a path between chemicals detected by mass spectrometry to original substances 42
44. 44. 43
45. 45. MS-Ready Mappings from Details Page 44
46. 46. MS-Ready Mappings Set of 20 substances for “PFOS” 45
47. 47. Mass and Formula Searching 46
48. 48. Advanced Searches Mass Search 47
49. 49. Advanced Searches Mass Search 48
50. 50. MS-Ready Structures for Formula Search 49
51. 51. MS-Ready Mappings • EXACT Formula: C10H16N2O8: 3 Hits 50
52. 52. MS-Ready Mappings • Same Input Formula: C10H16N2O8 • MS Ready Formula Search: 125 Chemicals 51
53. 53. MS-Ready Mappings • 125 chemicals returned in total – 8 of the 125 are single component chemicals – 3 of the 8 are isotope-labeled – 3 are neutral compounds and 2 are charged • Multiple components, stereo, isotopes and charge all collapsed and mapped through MS-Ready 52
54. 54. Batch Searching mass and formula 53
55. 55. Batch Searching • Singleton searches are useful but we work with thousands of masses and formulae! • Typical questions – What is the list of chemicals for the formula CxHyOz – What is the list of chemicals for a mass +/- error – Can I get chemical lists in Excel files? In SDF files? – Can I include properties in the download file? 54
56. 56. Batch Searching Formula/Mass 55
57. 57. Searching batches using MS-Ready Formula (or mass) searching 56
58. 58. Batch Search in specific lists 57
59. 59. Benefits of bringing it all together • The true dashboard benefit is integration • Rank potential candidates for toxicity using available data – hazard, exposure, in vitro 58
60. 60. Candidate ranking using metadata 59
61. 61. Data Source Ranking of “known unknowns” 60 • A mass and/or formula search is for an unknown chemical but it is a known chemical contained within a reference database • Most likely candidate chemicals have the most associated data sources, most associated literature articles or both C14H22N2O3 266.16304 Chemical Reference Database Sorted candidate structures
62. 62. Data Streams for Ranking • CompTox Dashboard Data Sources • PubChem Data Source Count • PubMed Reference Count • Toxcast in vitro bioactivity • Presence in CPDat database • OPERA PhysChem Properties • Other possibilities – predicted media occurrence, frequency of InChIs online
63. 63. Search 228.115 +/- 5.0 ppm 234 single component chemicals 62
64. 64. Search 228.115 +/- 5.0 ppm 234 single component chemicals 63
65. 65. The original ChemSpider work 64
66. 66. Is a bigger database better? 65 • ChemSpider was 26 million chemicals for the original work • Much BIGGER today • Is bigger better?? • Are there other metadata to use for ranking?
67. 67. Comparing Search Performance 66 • When dashboard contained 720k chemicals • Only 3% of ChemSpider size • What was the comparison in performance?
68. 68. SAME dataset for comparison 67
69. 69. How did performance compare? 68 For the same 162 chemicals, Dashboard outperforms ChemSpider for both Mass and Formula Ranking
70. 70. Identification ranks for 1783 chemicals using multiple data streams 69 DS: Data Sources PC: PubChem PM: PubMed STOFF: DB KEMI: DB Data Sources alone rank ~75% of the chemicals as Top Hit
71. 71. “UVCB” Chemicals 70
72. 72. UVCB Chemicals 71
73. 73. UVCBs challenge in non-target analysis 72 Homologue screening plots from Swiss Wastewater (Schymanski et al 2014, left) and Novi Sad (right) o Complex mixtures (UVCBs) are a huge and very challenging part of the unknowns in many environmental samples
74. 74. Public TSCA Inventory on Dashboard 31,460 Chemicals (1/24/2020) 73
75. 75. Many Chemicals are “Complex” >14000 chemicals are UVCBs 74
76. 76. “Markush Structures” https://en.wikipedia.org/wiki/Markush_structure 75
77. 77. How to represent complexity? 76
78. 78. In the Dashboard Abstract Sifter 77
79. 79. Literature Searching 78
80. 80. Literature Searching 79
81. 81. Abstract Sifter for Excel 80
82. 82. Work in Progress 81
83. 83. List Registration Activities • Registering and curating numerous lists – NIST library of chemicals –clean up especially around stereochemical representation – United States Geological Survey chemicals in water – Scientific Working Group for the Analysis of Seized Drugs – Synthetic Cannabinoids – Blood Exposome Database 82
84. 84. Blood Exposome Curation 83 • Blood exposome data collection from Barupal and Fiehn. Great work and we reviewing. • Aggregating large datasets is CHALLENGING • Comparing with our “Abstract Sifter” approach • We will iterate into a dashboard form..
85. 85. Prototype Work in Progress • CFM-ID – Viewing and Downloading pre-predicted spectra – Search spectra against the database • Structure/substructure/similarity search • Access to API and web services 84
86. 86. Predicted Mass Spectra http://cfmid.wishartlab.com/ • MS/MS spectra prediction for ESI+, ESI-, and EI • Predictions generated and stored for >800,000 structures, to be accessible via Dashboard 85
87. 87. Search Expt. vs. Predicted Spectra
88. 88. Search Expt. vs. Predicted Spectra
89. 89. Spectral Viewer Comparison 88
90. 90. Predicted Data Already Public Publication and Data Files 89 https://epa.figshare.com/articles/CFM-ID_Paper_Data/7776212/1
91. 91. Published: Chao et al 90
92. 92. Prototype Development 91
93. 93. CASMI 2012-2017 revisited • Application of metadata candidate ranking and CFM-ID to all five years of CASMI data 92
94. 94. Method Amenability Prediction Charlie Lowe Why? • Chromatography-mass spectrometry can be LC or GC • Which phase is more appropriate for which chemicals?
95. 95. Ongoing Work • Data sources to date • Massbank of North America • 9,275 chemicals for non-derivatized GC • 846 chemicals for derivatized GC • 816 chemicals for APCI+ • 454 chemicals for APCI- • 4,907 chemicals for ESI+ • 3,430 chemicals for ESI- • EPA Non-targeted Analysis Collaborative Trial (ENTACT) • 886 chemicals for non-derivatized GC • 44 chemicals for derivatized GC • 774 chemicals for APCI+ • 431 chemicals for APCI- • 1,113 chemicals for ESI+ • 648 chemicals for ESI-
96. 96. TMAP Visualization of MoNA GC Data
97. 97. Future Work: Add database of Collision Cross Section Prediction 96
98. 98. API services and Open Data • Web Services https://actorws.epa.gov/actorws/ • Data sets also available for download.. 97
99. 99. Web Services https://actorws.epa.gov/actorws/ • Data in UI, JSON and XML format • Our services are free of course.. 98
100. 100. InChIKey to DTXCIDs 99 https://actorws.epa.gov/actorws/dsstox/v02/msready?identifier =UVOFGKIRTCCNKG-UHFFFAOYSA-N
101. 101. Data and Services used by the Community 100
102. 102. NORMAN Suspect List Exchange https://www.norman-network.com/?q=node/236 101
103. 103. Integration to MetFrag in place https://jcheminf.biomedcentral.com/articles/10.1186/s13321-018-0299-2 102
104. 104. MassBank mapping to Dashboard Based on Web Service lookup 103
105. 105. Conclusion • Dashboard access to data for ~875,000 chemicals (~895k in the Spring Release) • MS-Ready data facilitates structure identification • Related metadata facilitates candidate ranking 104 • Relationship mappings and chemical lists of great utility • Curation and mutual sharing of chemical lists is important (e.g. NORMAN)
106. 106. ILS Kamel Mansouri EPA ORD Ann Richard Chris Grulke Jeremy Dunne Jeff Edwards Grace Patlewicz Alex Chao Kristin Isaacs Charles Lowe James McCord Seth Newton Katherine Phillips Jon Sobus Mark Strynar Elin Ulrich Joach Pleil GDIT Ilya Balabin Tom Transue Tommy Cathey Acknowledgements TEAMS IT Development Team Curation Team Collaborators Emma Schymanski NORMAN Network Andrew McEachran
107. 107. MANY presentations online https://tinyurl.com/w5hqs55 106
108. 108. Contact Antony Williams CCTE, US EPA Office of Research and Development, Williams.Antony@epa.gov ORCID: https://orcid.org/0000-0002-2668-4821 107 https://doi.org/10.1186/s13321-017-0247-6