IDDkin: network-based influence deep diffusion model for enhancing prediction of kinase inhibitors

Area of Science:

• Computational biology
• Drug discovery and development
• Bioinformatics

Background:

• Protein kinases are crucial drug targets due to their role in human diseases.
• Traditional kinase inhibitor prediction methods are inefficient and time-consuming.
• Network pharmacology and computational methods offer promising alternatives for drug discovery.

Purpose of the Study:

• To develop and validate a novel network-based computational model for predicting kinase inhibitors.
• To enhance the efficiency and accuracy of identifying potential kinase-targeting drugs.

Main Methods:

• Proposed an influence deep diffusion model (IDDkin) utilizing deep graph convolutional networks, graph attention networks, and adaptive weighting.
• Employed heterogeneous network information diffusion to update kinase and compound representations.
• Predicted potential compound-kinase interactions based on updated representations.

Main Results:

• IDDkin demonstrated superior performance compared to state-of-the-art and classic prediction methods.
• The model showed excellent generalizability and performance in case studies.
• Validated the powerful predictive ability of IDDkin in the kinase inhibitor domain.

Conclusions:

• IDDkin offers a highly effective and accurate computational approach for kinase inhibitor prediction.
• The model advances network-based computational strategies in drug discovery.
• IDDkin provides a valuable tool for accelerating the development of novel kinase-targeting therapies.