A Novel Graph Neural Network Methodology to Investigate Dihydroorotate Dehydrogenase Inhibitors in Small Cell Lung...

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Small cell lung cancer (SCLC) is an aggressive malignancy with limited therapeutic options.
• Dihydroorotate dehydrogenase (DHODH) is a validated therapeutic target in SCLC.
• Novel strategies are needed to identify potent and selective DHODH inhibitors.

Purpose of the Study:

• To identify potential DHODH inhibitors for SCLC treatment.
• To develop and validate advanced computational models for drug discovery.
• To explore the utility of multi-graph neural networks (multi-GNNs) in predicting drug efficacy.

Main Methods:

• Network pharmacology and virtual screening were employed to identify potential drug targets and candidates.
• Three novel multi-GNN models (GIAN, GIAT, SGCA) were developed for predictive modeling.
• Machine learning algorithms including random forest (RF) and support vector regression (SVR) were used for comparison.
• Molecular dynamics simulations were performed to assess the stability of protein-ligand complexes.

Main Results:

• The developed multi-GNN models demonstrated high accuracy, with R² values > 0.92 on the training set and > 0.8 on the test set.
• Multi-GNNs outperformed traditional machine learning methods in modeling effect and precision.
• Molecular dynamics simulations confirmed the stability of identified protein-ligand complexes.
• ZINC8577218, ZINC95618747, and ZINC4261765 were identified as potential potent DHODH inhibitors.

Conclusions:

• Multi-GNNs represent a powerful and precise approach for drug discovery, particularly for identifying inhibitors of therapeutic targets like DHODH in SCLC.
• The novel multi-GNN framework offers a promising protocol for future drug discovery endeavors.
• The identified compounds warrant further investigation as potential therapeutic agents for SCLC.