Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dumpling GNN: Hybrid GNN Enables Better ADC Payload Activity Prediction Based on Chemical Structure

Shengjie Xu, Lingxi Xie

TL;DR

DumplingGNN represents a significant advancement in molecular property prediction, with particular promise for accelerating the design and optimization of ADC payloads in targeted cancer therapy development.

Abstract

Antibody-drug conjugates (ADCs) have emerged as a promising class of targeted cancer therapeutics, but the design and optimization of their cytotoxic payloads remain challenging. This study introduces DumplingGNN, a novel hybrid Graph Neural Network architecture specifically designed for predicting ADC payload activity based on chemical structure. By integrating Message Passing Neural Networks (MPNN), Graph Attention Networks (GAT), and GraphSAGE layers, DumplingGNN effectively captures multi-scale molecular features and leverages both 2D topological and 3D structural information. We evaluate DumplingGNN on a comprehensive ADC payload dataset focusing on DNA Topoisomerase I inhibitors, as well as on multiple public benchmarks from MoleculeNet. DumplingGNN achieves state-of-the-art performance across several datasets, including BBBP (96.4\% ROC-AUC), ToxCast (78.2\% ROC-AUC), and PCBA (88.87\% ROC-AUC). On our specialized ADC payload dataset, it demonstrates exceptional accuracy (91.48\%), sensitivity (95.08\%), and specificity (97.54\%). Ablation studies confirm the synergistic effects of the hybrid architecture and the critical role of 3D structural information in enhancing predictive accuracy. The model's strong interpretability, enabled by attention mechanisms, provides valuable insights into structure-activity relationships. DumplingGNN represents a significant advancement in molecular property prediction, with particular promise for accelerating the design and optimization of ADC payloads in targeted cancer therapy development.

Dumpling GNN: Hybrid GNN Enables Better ADC Payload Activity Prediction Based on Chemical Structure

TL;DR

DumplingGNN represents a significant advancement in molecular property prediction, with particular promise for accelerating the design and optimization of ADC payloads in targeted cancer therapy development.

Abstract

Antibody-drug conjugates (ADCs) have emerged as a promising class of targeted cancer therapeutics, but the design and optimization of their cytotoxic payloads remain challenging. This study introduces DumplingGNN, a novel hybrid Graph Neural Network architecture specifically designed for predicting ADC payload activity based on chemical structure. By integrating Message Passing Neural Networks (MPNN), Graph Attention Networks (GAT), and GraphSAGE layers, DumplingGNN effectively captures multi-scale molecular features and leverages both 2D topological and 3D structural information. We evaluate DumplingGNN on a comprehensive ADC payload dataset focusing on DNA Topoisomerase I inhibitors, as well as on multiple public benchmarks from MoleculeNet. DumplingGNN achieves state-of-the-art performance across several datasets, including BBBP (96.4\% ROC-AUC), ToxCast (78.2\% ROC-AUC), and PCBA (88.87\% ROC-AUC). On our specialized ADC payload dataset, it demonstrates exceptional accuracy (91.48\%), sensitivity (95.08\%), and specificity (97.54\%). Ablation studies confirm the synergistic effects of the hybrid architecture and the critical role of 3D structural information in enhancing predictive accuracy. The model's strong interpretability, enabled by attention mechanisms, provides valuable insights into structure-activity relationships. DumplingGNN represents a significant advancement in molecular property prediction, with particular promise for accelerating the design and optimization of ADC payloads in targeted cancer therapy development.
Paper Structure (40 sections, 5 equations, 4 figures, 4 tables)

This paper contains 40 sections, 5 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Antibody-Drug Conjugates (ADCs) and Payload Importance
  3. Machine Learning in Molecular Property Prediction
  4. Graph Neural Networks: Advancements and Challenges in Molecular Modeling
  5. Introducing DumplingGNN
  6. Main Contributions
  7. Paper Organization
  8. Related work
  9. Graph Neural Networks in Drug Discovery
  10. Hybrid GNN Architectures
  11. ADC Payload Activity Prediction
  12. Incorporation of 3D Structural Information
  13. Interpretability in Molecular GNNs
  14. Limitations of Existing Approaches
  15. Method
  16. ...and 25 more sections

Figures (4)

  • Figure 1: Workflow of DumplingGNN architecture for ADC payload activity prediction. The hybrid architecture integrates MPNN, GAT, and GraphSAGE layers to capture multi-scale molecular features.
  • Figure 2: The process of message computation and node update in MPNN. This mechanism allows for the modeling of local chemical interactions.
  • Figure 3: Graph Attention Network (GAT) in DumplingGNN. The attention mechanism allows the model to focus on the most relevant atomic interactions for activity prediction, enhancing both performance and interpretability.
  • Figure 4: GraphSAGE layer in DumplingGNN. This layer aggregates information across different scales, capturing global molecular properties.