How Open and FAIR Cheminformatics can Support the Discovery, Analysis and Assessment of PFAS
Authors/Creators
- 1. LCSB, Uni Luxembourg
- 2. NCBI/NLM/NIH
- 3. IBED, Uni Amsterdam
Description
Presentation in the " Poly- and perfluoroalkyl substances (PFASs): Addressing Urgent Questions in the 21st Century" session in the "Environmental chemistry and exposure assessment: analysis, monitoring, fate and modelling" track at SETAC Europe 2022 in Copenhagen, May 15-19, 2022.
The PubChem PFAS Tree can be found here: https://pubchem.ncbi.nlm.nih.gov/classification/#hid=120.
The documentation can be found here: https://gitlab.lcsb.uni.lu/eci/pubchem-docs/-/raw/main/pfas-tree/PFAS_Tree.pdf?inline=false
Abstract:
The ever-increasing concerns about poly and perfluoroalkyl substances (PFAS) and calls for action upon them as a class has spurred intense debates on how to define and enumerate the “PFAS Chemical Space”. There are now a number of PFAS lists openly available to the community. These include the OECD PFAS list of ~4700 PFAS (ENV/JM/MONO(2018)7), the US EPA PFASMASTER list of ~9000 PFAS. Over 30 more are available on resources such as the CompTox Chemicals Dashboard (https://comptox.epa.gov/dashboard/chemical_lists/?search=PFAS) and the NORMAN Suspect List Exchange (https://zenodo.org/communities/norman-sle/search?page=1&size=20&q=PFAS). However, searching the large open chemical collection PubChem (https://pubchem.ncbi.nlm.nih.gov/; 111 million chemicals) reveals that ~6 million entries match the latest OECD PFAS definition where PFAS “contains at least one alkyl CF2 group”. Navigating this space is challenging, between the tens of PFAS routinely monitored, thousands on existing lists, and the millions existing in databases. This work demonstrates how the cross-resource integration of information between the NORMAN Suspect List Exchange and PubChem can facilitate closing the gaps in major open databases and fast-track publication of new structures identified in research (e.g., PFAS transformation products; previously undocumented PFAS) including detailed annotations such as reactions, spectral fragmentation patterns, collision cross section values and even NMR spectra. By highlighting and building on existing PubChem functionality, this in turn offers extensive bulk property calculations, patent information, available annotation (such as: uses, toxicity, and biomedical literature), download files, and advanced programmatic access. Since discovery of PFAS in the environment is critical, a relevant subset of PFAS in PubChem (~255K chemicals with –CF2CF2–) is made available as a PFAS database for non-target or suspect screening mass spectrometry efforts for incorporation into open software such as patRoon (https://rickhelmus.github.io/patRoon/), MetFrag (https://msbi.ipb-halle.de/MetFrag/) and more. A set of findable, accessible, interoperable, reusable (FAIR) templates are provided to enhance data sharing between resources and increase the availability of FAIR datasets for future data-hungry computational developments such as literature/patent mining, transformation product prediction (e.g. BioTransformer) and/or automated PFAS classification efforts.