PFAS Rigidity
Modeling.
Applying Hybrid GNN architectures to predict the structural rigidity and environmental persistence of Per- and Polyfluoroalkyl Substances in industrial water cycles.
The Incentive
Predicting degradation
before synthesis.
Industrial water treatment systems are reactive. Our model allows manufacturers to predict the 100-year environmental persistence of a chemical surfactant before it even leaves the simulation environment.
The Methodology
Hybrid GNN Architecture.
Combining Graph Convolutional Networks (GCN) with RDKit molecular descriptors to map toxicity across 1.2M known compounds. Transfer learning from ChEMBL ensures zero-shot prediction on fluorinated chains.
Data Pipeline
Automated ingestion of Tox21 & ToxCast datasets with structural validation via RDKit.
Edge Compute
Model optimized for inference at the water filtration system using Nvidia Jetson.
Environmental Impact
Predicting half-life degradation across 4 major industrial effluent types.
Research Outcomes
Derived
Discovery.
14 Bio-degradable Surfactant leads identified.
60% faster environmental half-life predicted.
Acquisition Point
This research module is available for exclusive commercial licensing or integration into proprietary R&D pipelines.