GNN-DDAS: Drug discovery for identifying anti-schistosome small molecules based on graph neural network

Area of Science:

• Computational chemistry
• Drug discovery
• Machine learning

Background:

• Schistosomiasis is a neglected tropical disease affecting millions.
• Praziquantel, the primary treatment, faces challenges like drug resistance and limited efficacy in children.
• Accurate computer-aided methods are needed to discover new anti-schistosome drugs.

Purpose of the Study:

• To develop a novel deep learning framework, GNN-DDAS, for identifying active anti-schistosome small molecules.
• To improve the accuracy of computer-aided drug discovery for schistosomiasis.

Main Methods:

• Utilized a multi-layer perceptron for sequence feature extraction from small molecule SMILES.
• Employed graph neural networks (GNN) to extract structural features from molecular graphs.
• Integrated sequence and structural features into a fully connected network for prediction.

Main Results:

• GNN-DDAS demonstrated superior performance over benchmark methods on both benchmark and real-world datasets.
• The GNNExplainer model provided insights into key substructure features driving drug activity.
• The framework successfully predicted active anti-schistosome small molecules.

Conclusions:

• GNN-DDAS offers a promising and accurate computational approach for discovering new anti-schistosome drugs.
• The framework enhances drug discovery efficiency by analyzing both sequence and structural molecular features.
• This method addresses critical unmet needs in schistosomiasis treatment development.