Drug repositioning model based on knowledge graph embedding.

Area of Science:

• Computational biology
• Pharmacology
• Data science

Background:

• Drug repositioning accelerates therapeutic development by repurposing existing drugs.
• Integrating complex, scattered data into a cohesive knowledge network is a significant challenge.
• Existing methods struggle with the scale and heterogeneity of biomedical data.

Purpose of the Study:

• To develop an advanced drug repositioning model using knowledge graph embedding.
• To enhance the accuracy and efficiency of identifying novel drug indications.
• To address the limitations of current data integration and knowledge organization in drug discovery.

Main Methods:

• Proposed a knowledge graph embedding model utilizing multivariate relational data.
• Developed novel attention-based models (Attranse, Attdismult, Attrescal) by integrating attention mechanisms into translation and bilinear models.
• Employed a multi-model feature extraction approach, combining results for improved drug screening.

Main Results:

• The model demonstrated high accuracy in identifying new drug indications, validated using COVID-19 data.
• Successfully identified 7 clinically approved drugs for COVID-19 treatment, aligning with existing clinical applications.
• The integrated approach significantly enhanced drug screening quality compared to single-model methods.

Conclusions:

• The proposed knowledge graph embedding model offers a powerful tool for drug repositioning.
• The model's effectiveness in identifying COVID-19 treatments highlights its potential for emerging infectious diseases.
• This approach provides valuable insights for accelerating drug development for complex conditions.