Abstract:Per- and polyfluoroalkyl substances (PFAS) are persistent environmental pollutants with known toxicity and bioaccumulation issues. Their widespread industrial use and resistance to degradation have led to global environmental contamination and significant health concerns. While a minority of PFAS have been extensively studied, the toxicity of many PFAS remains poorly understood due to limited direct toxicological data. This study advances the predictive modeling of PFAS toxicity by combining semi-supervised graph convolutional networks (GCNs) with molecular descriptors and fingerprints. We propose a novel approach to enhance the prediction of PFAS binding affinities by isolating molecular fingerprints to construct graphs where then descriptors are set as the node features. This approach specifically captures the structural, physicochemical, and topological features of PFAS without overfitting due to an abundance of features. Unsupervised clustering then identifies representative compounds for detailed binding studies. Our results provide a more accurate ability to estimate PFAS hepatotoxicity to provide guidance in chemical discovery of new PFAS and the development of new safety regulations.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of the hepatotoxicity mechanism of per - and polyfluoroalkyl substances (PFAS). Specifically, the research objective is to predict the binding affinity between PFAS and liver fatty - acid - binding protein (L - FABP) by combining graph convolutional network (GCN) and computational modeling methods, and to reveal its underlying toxicity mechanism. #### Main problem background 1. **Widespread pollution and health risks of PFAS**: - PFAS are a class of persistent environmental pollutants with known toxicity and bioaccumulation problems. - Their wide - spread use and resistance to degradation have led to global environmental pollution and significant health concerns. - Although some PFAS have been widely studied, the toxicity of many PFAS remains unclear, lacking direct toxicological data. 2. **Limitations of existing research**: - At present, most research focuses on a few common PFAS, such as perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), while relatively little research has been done on other PFAS. - Due to the low concentration of PFAS in biological samples, it is difficult to conduct direct toxicological experiments. - Epidemiological studies have shown that PFAS are associated with multiple health effects, especially on liver and lipid metabolism. 3. **L - FABP as a key target**: - Liver fatty - acid - binding protein (L - FABP) is an important PFAS target because PFAS are structurally similar to natural fatty acids, which may lead to bioaccumulation in these proteins. - Studies have shown that the stronger the binding affinity between PFAS and L - FABP, the stronger their toxicity usually is. #### Methods proposed in the paper To overcome the above challenges, this paper proposes the following methods: 1. **Combining GCN and molecular descriptors/fingerprints**: - Use a semi - supervised graph convolutional network (GCN) model to capture the structural, physicochemical, and topological features of PFAS through chemical fingerprints and descriptors. - This method can avoid over - fitting problems caused by too many features and improve prediction accuracy at the same time. 2. **Unsupervised clustering to select representative compounds**: - Identify representative PFAS compounds through unsupervised clustering methods (such as Ward's algorithm) for detailed binding studies. - This helps to reduce the amount of data and improve computational efficiency. 3. **Molecular dynamics simulation to reveal binding mechanisms**: - Conduct molecular dynamics (MD) simulations on selected PFAS to gain in - depth understanding of their specific binding mechanisms with L - FABP. #### Significance of the research This research not only provides more accurate PFAS hepatotoxicity prediction capabilities but also provides guidance for the chemical discovery of new PFAS and the formulation of new safety regulations. Through this method, researchers can better understand the toxicity mechanism of PFAS, thereby providing a scientific basis for environmental protection and human health. ### Summary This paper attempts to solve the problems of PFAS hepatotoxicity prediction and mechanism revelation by combining advanced technologies such as graph convolutional network (GCN) and molecular dynamics simulation, especially to improve the accuracy and reliability of prediction under limited toxicological data.

Uncovering the Mechanism of Hepatotoxiciy of PFAS Targeting L-FABP Using GCN and Computational Modeling

Hepatotoxicity Assessment Investigations on PFASs Targeting L-FABP Using Binding Affinity Data and Machine Learning-Based QSAR Model.

Investigation of the Binding Fraction of PFAS in Human Plasma and Underlying Mechanisms Based on Machine Learning and Molecular Dynamics Simulation

Molecular designing of potential environmentally friendly PFAS based on deep learning and generative models

Exploring toxicity of perfluorinated compounds through complex network and pathway modeling

Prediction of the Interactions of a Large Number of Per- and Poly-Fluoroalkyl Substances with Ten Nuclear Receptors

Elucidation of the Molecular Determinants for Optimal Perfluorooctanesulfonate Adsorption Using a Combinatorial Nanoparticle Library Approach.

Identifying priority PBT-like compounds from emerging PFAS by nontargeted analysis and machine learning models

Investigating mouse hepatic lipidome dysregulation following exposure to emerging per- and polyfluoroalkyl substances (PFAS)

Bioaccumulation Mechanisms of Perfluoroalkyl Substances (pfass) in Aquatic Environments: Theoretical and Experimental Insights

Elucidating Key Characteristics of PFAS Binding to Human Peroxisome Proliferator-Activated Receptor Alpha: An Explainable Machine Learning Approach

Integrating high-throughput phenotypic profiling and transcriptomic analyses to predict the hepatosteatosis effects induced by per- and polyfluoroalkyl substances

Category-Based Toxicokinetic Evaluations of Data-Poor Per- and Polyfluoroalkyl Substances (PFAS) using Gas Chromatography Coupled with Mass Spectrometry

Screening of Potential PFOS Alternatives To Decrease Liver Bioaccumulation: Experimental and Computational Approaches.

Insights into the Competitive Mechanisms of Per- and Polyfluoroalkyl Substances Partition in Liver and Blood

Effect of Enterohepatic Circulation on the Accumulation of Per- and Polyfluoroalkyl Substances: Evidence from Experimental and Computational Studies

A novel molecular pathway of lipid accumulation in human hepatocytes caused by PFOA and PFOS

Disclosing the bioaccumulation and biomagnification behaviors of emerging per/polyfluoroalkyl substances in aquatic food web based on field investigation and model simulation

Characterizing PFASs in aquatic ecosystems with 3D hydrodynamic and water quality models

Microfluidic Chip-Based Modeling of Three-Dimensional Intestine-Vessel-Liver Interactions in Fluorotelomer Alcohol Biotransformation

Advancing Toxicity Studies of Per- and Poly-Fluoroalkyl Substances (pfass) Through Machine Learning: Models, Mechanisms, and Future Directions