An artificial intelligence platform for automated PFAS subgroup classification: A discovery tool for PFAS screening
An Su1, Yingying Cheng1, Chengwei Zhang2
1State Key Laboratory Breeding Base of Green Chemistry-Synthesis Technology, Key Laboratory of Green Chemistry-Synthesis Technology of Zhejiang Province, College of Chemical Engineering, Zhejiang University of Technology, Hangzhou, Zhejiang 310014, China; Key Laboratory of Pharmaceutical Engineering of Zhejiang Province, Collaborative Innovation Center of Yangtze River Delta Region Green Pharmaceuticals, Zhejiang University of Technology, Hangzhou, Zhejiang 310014, PR China.
A new AI platform, PFAS-Atlas, categorizes and groups per- and polyfluoroalkyl substances (PFAS) more effectively. It aids researchers and regulators by visualizing PFAS chemical space and guiding testing strategies.
Area of Science:
- Environmental Chemistry
- Computational Chemistry
- Toxicology
Background:
- Rapid discovery of per- and polyfluoroalkyl substances (PFAS) outpaces structural analysis and toxicity assessment.
- Effective categorization and grouping of PFAS are urgently needed for research and regulation.
Purpose of the Study:
- To introduce PFAS-Atlas, an AI-based platform for classifying and grouping PFAS.
- To improve upon existing software by adhering to the latest OECD definition of PFAS and minimizing uncategorized substances.
- To visualize the chemical space of PFAS and identify relationships between chemical structure and potential risks.
Main Methods:
- Development of a rule-based automatic classification system for PFAS.
- Implementation of a machine learning-based grouping model using deep unsupervised learning.
- Clustering of similar PFAS structures and linking of related chemical classes.
- Application of the platform to real-world use cases for screening and strategy planning.
Main Results:
- PFAS-Atlas successfully classifies PFAS according to the latest OECD definition, reducing uncategorized compounds.
- Deep unsupervised learning visualizes the PFAS chemical space, revealing clusters of similar structures.
- The platform demonstrates rapid screening of structure-activity relationships for persistence, bioaccumulation, and toxicity.
- PFAS-Atlas effectively guides PFAS testing strategies by highlighting areas needing further investigation.
Conclusions:
- PFAS-Atlas provides a powerful tool for the classification and grouping of PFAS.
- The platform enhances understanding of PFAS chemical space and structure-activity relationships.
- PFAS-Atlas will significantly benefit the PFAS research and regulatory communities by improving efficiency and informing testing strategies.
More Related Videos
16:02Demonstration of the Sequence Alignment to Predict Across Species Susceptibility Tool for Rapid Assessment of Protein Conservation
Published on: February 10, 2023
07:06Investigating Long-Distance Transport of Perfluoroalkyl Acids in Wheat via a Split-Root Exposure Technique
Published on: September 28, 2022
