Advancing drug-target interaction prediction: a comprehensive graph-based approach integrating knowledge graph...

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Drug target interaction (DTI) validation is costly and time-consuming, limiting experimental verification.
• Accurate computational methods are crucial for accelerating drug discovery and predicting potential DTIs.
• Graph-based machine learning and knowledge graph embedding (KGE) are emerging as powerful tools for DTI prediction.

Purpose of the Study:

• To develop and evaluate a novel computational approach, DTIOG, for predicting drug-target interactions (DTIs).
• To leverage knowledge graph embedding and contextual protein sequence information for enhanced DTI prediction accuracy.

Main Methods:

• DTIOG employs a two-step knowledge graph embedding (KGE) process to compute embedding vectors.
• Protein Bidirectional Encoder Representations from Transformers (ProtBERT) is utilized to derive contextual information from protein sequences and assess target-target similarity.
• The approach integrates local representations from drug SMILES strings and protein amino acid sequences.

Main Results:

• DTIOG demonstrated high efficacy in predicting DTIs across datasets for Enzymes, Ion Channels, and G-protein-coupled Receptors.
• The method outperformed existing algorithms in DTI prediction accuracy when utilizing various similarity measures and classifiers.
• The robustness and accuracy of DTIOG were consistently observed across all tested datasets.

Conclusions:

• DTIOG is a robust and accurate method for predicting drug-target interactions.
• The proposed approach shows significant potential for accelerating drug discovery by identifying novel DTIs.
• DTIOG can serve as a valuable tool for researchers in the pharmaceutical field.