[SGMS 2025] Enhancing Interpretation of Non-target MS data with Open Chemical Knowledge

Prof. Dr. Emma L. Schymanski (plus many, many colleagues and collaborators!) Environmental Cheminformatics Group, Luxembourg Centre for Systems Biomedicine, University of Luxembourg [email protected] / https://gitlab.com/uniluxembourg/lcsb/eci https://wwwen.uni.lu/lcsb/research/environmental_cheminformatics/ Enhancing Interpretation of Non-target MS data with Open Chemical Knowledge Slides available at DOI: 10.5281/zenodo.17390447 42nd Swiss Group for Mass Spectrometry (SGMS) Meeting Beatenberg, 23-24 October 2025 Environmental Cheminformatics @ LCSB, Uni.lu FAIR chemical data Open source mass spectral library HRMS NTS TP workflows Slide inspired by B. Talavera Explore chemical space Browse transformations 2 Disease impacts 122 million The Vast Chemical Space of Exposomics … Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ 122 million The Vast Chemical Space of Exposomics … Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ https://pubchem.ncbi.nlm.nih.gov/ 122 million (Unfortunately, our major open resources are under threat) Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ https://pubchem.ncbi.nlm.nih.gov/ 122 million < 0.01% of PubChem The Vast Chemical Space of Exposomics … Target Analysis Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ 122 million < 0.01% of PubChem The Vast Chemical Space of Exposomics … Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ https://massbank.eu/MassBank/ https://github.com/MassBank/MassBank-data/ Neumann et al. (accepted) Nucleic Acid Research Database Issue MassBank.EU – Environmental Data Contributions 8 https://massbank.eu/MassBank/content 122 million < 0.01% of PubChem The Vast Chemical Space of Exposomics … 133,573 suspects Suspect Screening Target Analysis Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ Suspect List(s) (NORMAN-SLE) (NORMAN SusDat) NORMAN Database System (NDS) Mohammed Taha et al. (2022) DOI: 10.1186/s12 302-02200680-6 16 Mohammed Taha et al. (2022) DOI: 10.1186/s12302-022-00680-6 17 122 million < 0.01% of PubChem The Vast Chemical Space of Exposomics … 566,122 133,573 suspects Non-target Screening Suspect Screening Target Analysis Numbers updated 9 Oct. 2025. Mod. from Escher et al (2020). Science. DOI: 10.1126/science.aay6636 Image: ESO/IDA/Danish 1.5 m http://www.eso.org/public/images/potw1015a/ Bottleneck: Data Interpretation and Generating Insights 19 Mod. From Fig. 2, Sevin et al 2015, Current Opinion in Biotechnology. DOI: 10.1016/j.copbio.2014.10.001 Grand Challenge: HR-MS “Chemical Space” is too Big! 290 million 122 million 129 million Candidates with high information content Candidates with low information content 1.2 million …and prohibitive run times! 20 Creating Informative Subsets of 21 Elapavalore et al. (2025) DOI: 10.1021/acs.estlett.4c01003 Schymanski et al. (2021) DOI: 10.1186/s13321-021-00489-0 Creating Informative Subsets of ~500,000 entries “small” Elapavalore et al. (2025) DOI: 10.1021/acs.estlett.4c01003 Schymanski et al. (2021) DOI: 10.1186/s13321-021-00489-0 22 for Exposomics 23 http://pubchemlite.lcsb.uni.lu/ caffeine Database: for Exposomics 24 https://pubchemlite.lcsb.uni.lu/e/compound/2519 caffeine Database: for Exposomics 25 https://pubchemlite.lcsb.uni.lu/e/compound/2519 caffeine Time Trend Analysis Reveals Behaviour Change 32 Aurich et al. (2024) ES&T Letters DOI: 10.1021/acs.estlett.4c00560 https://gitlab.com/uniluxembourg/lcsb/eci/chemicalstripes The Chemical Stripes and Patent Data Left: Neuwald et al, STOTEN, DOI: 10.1016/j.scitotenv.2023.163921 Photo of Daniel Zahn, UFZ at SETAC Europe, 30 April – 4 May, 2023. Image reused with permission EV 33 (live stripe calculations during SETAC 2023) …and now in 2025: Fluorinated ions are everywhere… Tisler et al. (2025) DOI: 10.1021/acs.est.5c06876 34 Network Analysis to find Patent Trends 35 Aurich et al. (2024) ES&T Letters DOI: 10.1021/acs.estlett.4c00560 https://gitlab.com/uniluxembourg/lcsb/eci/chemicalstripes Elapavalore et al. (2025) DOI: 10.1021/acs.estlett.4c01003 & Schymanski et al. (2021) DOI: 10.1186/s13321-021-00489-0 36 : Fewer and more relevant candidates – with context is a Dynamic Collection … •… built every week, updated online every month •… re-evaluated with every build for continuous quality control •… growing as knowledge grows! DOI: 10.5281/ zenodo.5995885 37 http://pubchemlite.lcsb.uni.lu/ Variation of Content Over Time (2022-2024) •Variation of content over time Elapavalore et al. (2025) DOI: 10.1021/acs.estlett.4c01003 38 Coverage of Content 39 Source: Talavera Andújar & Schymanski, under review, Exposome. https://gitlab.com/uniluxembourg/lcsb/eci/exposomics-plots. Triglycerides Nucleotides EU Pesticides PFOS and related substances How does perform? Schymanski et al. (2021) DOI: 10.1186/s13321-021-00489-0 40 18 April 2025 22 April 2025 + ChemClass Difference Total PCL Entries 410,755 442,233 31,478 Rank 1 793 (81.2 %) 797 (81.6 %) 4 Rank 1-2 916 (93.8 %) 917 (93.9 %) 1 Rank 1-5 961 (98.4 %) 961 (98.4 %) 0 Failures 11 (1.1 %) 10 (1.0 %) -1 Total Evaluated 977 (100 %) 977 (100 %) 0 File Size 171,518 190,197 9.82 % Elapavalore et al. (2025) DOI: 10.1021/acs.estlett.4c01003 >80 % Ranked first >93 % Ranked second => High throughput identification (quickly!) How does perform? Schymanski et al. (2021) DOI: 10.1186/s13321-021-00489-0 41 18 April 2025 22 April 2025 + ChemClass Difference Total PCL Entries 410,755 442,233 31,478 Rank 1 793 (81.2 %) 797 (81.6 %) 4 Rank 1-2 916 (93.8 %) 917 (93.9 %) 1 Rank 1-5 961 (98.4 %) 961 (98.4 %) 0 Failures 11 (1.1 %) 10 (1.0 %) -1 Total Evaluated 977 (100 %) 977 (100 %) 0 File Size 171,518 190,197 9.82 % Elapavalore et al. (2025) DOI: 10.1021/acs.estlett.4c01003 Is Regrettable Substitution Impacting Neurodegeneration? Talavera Andújar et al. (2024). Environment International DOI: 10.1016/j.envint.2024.109151 48 Control iPD PD-/LRRK2+ PD+/LRRK2+ -2 -1 0 1 2 BPS Normalized peak intensity p=0.0117 p=0.0591 p=0.0620 Emerged as a “safer” alternative to bisphenol A (BPA) Control iPD PD-/LRRK2+ PD+/LRRK2+ -3 -2 -1 0 1 2 3 PFBuS Normalized peak intensity p=0.0184 p=0.0069 p=0.0649 PFBuS is a substitute for perfluorooctanesulfonate (PFOS) 1,003 chemicals Perfluorobutane Sulfonic Acid Bisphenol S Use and manufacturing (549) Associated Disease/Disorder (543) Pathways (331) Hazardous chemical replacements (2) Compound TOC BPS Impacting Mitochondria in Patient-Derived Neurons Talavera Andújar et al. (2024). Environment International DOI: 10.1016/j.envint.2024.109151 49 Environmental Cheminformatics @ LCSB, Uni.lu FAIR chemical data Open source mass spectral library HRMS NTS TP workflows Slide inspired by B. Talavera Explore chemical space Browse transformations 50 Disease impacts Open Source Workflows for NT-HRMS: patRoon Helmus et al. (2022) patRoon 2.0 J. Open Source Software, DOI: 10.21105/joss.04029 Helmus et al. (2021) patRoon. J. Cheminformatics. DOI: 10.1186/s13321-020-00477-w 51 NTS in Luxembourg 52 Aurich et al (2023)Env. Sci. Europe. DOI: 10.1186/s12302-023-00805-5 Industrial chemicals OP flame retardants Transformation products time Applying TP NT-HRMS Workflows with patRoon 53 time Desethyl-atrazine TFA Benzidine TP library Suspect/Nontarget screening TP suspect screening Mod. from Palm et al (2023) ShinyTPs, DOI: 10.1021/acs.estlett.3c00537, Palm et al (2023). SETAC Presentation, DOI: 10.5281/zenodo.7885829 FAIR Transformations in PubChem and NORMAN-SLE Schymanski et al. (2021) DOI: 10.1186/s13321-021-00489-0; Helmus et al (2022) DOI: 10.21105/joss.04029 Schymanski, Bolton, Cheng, Thiessen, Zhang, Helmus (2021) Transformations in PubChem, DOI: 10.5281/zenodo.5644560 9,395 chemicals > 3x the BioTransformer library 54 Introducing FAIR-TPs 55 Mayahi, Palm & Schymanski (in prep.). https://fairtps.lcsb.uni.lu/stats/ & https://fairtps.lcsb.uni.lu/about/ Éiwegkeetschemikalie / polluants éternels / Forever chemicals? 56 https://www.rtl.lu/news/national/a/2242278.html / https://today.rtl.lu/news/luxembourg/a/2242399.html https://infos.rtl.lu/actu/luxembourg /a/2242403.html FAIR-TPs – Explore TPs Extracted From Literature 57 Mayahi, Palm & Schymanski (in prep.). https://fairtps.lcsb.uni.lu/ Predicting Dead-end Transformation Products 64 Source: Emma Palm. Palm et al. (2025) Presentation at SETAC Europe, 13 May 2025 . DOI: 10.5281/zenodo.15574837 Number of precursors: 13 Predicting Dead-end Transformation Products 65 Source: Emma Palm. Palm et al. (2025) Presentation at SETAC Europe, 13 May 2025 . DOI: 10.5281/zenodo.15574837 Number of precursors: 13 FAIR-TPs 66 Mayahi, Palm & Schymanski (in prep.). https://fairtps.lcsb.uni.lu/stats/ & https://fairtps.lcsb.uni.lu/about/ Environmental Cheminformatics @ LCSB, Uni.lu FAIR chemical data Open source mass spectral library HRMS NTS TP workflows Slide inspired by B. Talavera Explore chemical space Browse transformations 67 Disease impacts Email: emma.schymans[email protected] in:emma-schymanski / @[email protected] https://wwwen.uni.lu/lcsb/research/environmental_ cheminformatics/ https://gitlab.com/uniluxembourg/lcsb/eci/ https://pubchemlite.lcsb.uni.lu/ Merci/Danke schön/Thank you! H2020: 101036756 68