IIFDTI

IIFDTI predicts drug-target interactions using an end-to-end deep learning framework that integrates independent and interactive molecular features to support DTI discovery.

Key Features:

• End-to-End Deep Learning Framework: Integrates independent and interactive features of drug-target pairs to capture both individual and interaction-specific characteristics.
• Bidirectional Encoder-Decoder Architecture: Extracts interactive features of substructures between drugs and targets.
• Graph Neural Networks (GNNs) and Convolutional Neural Networks (CNNs): Uses GNNs to extract independent features from drug data and CNNs to extract independent features from target data.
• Feature Fusion and Dense Layer Integration: Fuses extracted features and processes them through fully connected dense layers for prediction.
• Attention Mechanism for Interpretability: Applies attention to provide visualizations that reveal biologically relevant interaction signals.

Scientific Applications:

• Benchmark Performance: Demonstrates higher area under the ROC curve (AUC), area under the precision-recall curve (AUPR), precision, and recall compared to state-of-the-art methods on benchmark datasets.
• Prioritization for Experimental Validation: Ranks and prioritizes candidate drug-target pairs to guide experimental verification and drug discovery efforts.
• Interpretability and Biological Insight: Attention-based visualizations reveal insights into the significance of specific interactions and substructures.
• Case Study Identification: Identifies promising drug-target pairs in case studies for further investigation.

Methodology:

End-to-end deep learning framework combining GNNs for drug independent features, CNNs for target independent features, a bidirectional encoder-decoder for interactive substructure features, feature fusion followed by fully connected dense layers for prediction, and an attention mechanism for interpretability.