Artificial intelligence to guide precision anticancer therapy with multitargeted kinase inhibitors

Area of Science:

• Bioinformatics and Computational Biology
• Oncology
• Artificial Intelligence in Medicine

Background:

• The convergence of big biological data and AI advancements is driving precision oncology.
• CancerOmicsNet is an AI system designed to predict the efficacy of multitargeted kinase inhibitors in cancer treatment.
• This AI approach surpasses existing methods like deep learning and molecular docking.

Purpose of the Study:

• To develop an AI-based system, CancerOmicsNet, for predicting therapeutic effects of kinase inhibitors.
• To enhance precision oncology through accurate and robust therapeutic predictions.
• To outperform current data-driven therapeutic discovery methods.

Main Methods:

• Integration of heterogeneous data using a deep graph learning model.
• Application of attention propagation mechanisms for feature extraction from cancer networks.
• Development of an AI system for superior prediction accuracy compared to gene signature reversion.

Main Results:

• CancerOmicsNet predictions validated against literature and live-cell assays for breast, pancreatic, and prostate cancers.
• Six molecules demonstrated dose-dependent antiproliferative effects.
• Pan-CDK inhibitor JNJ-7706621 and Src inhibitor PP1 showed high potency against pancreatic cancer cells.

Conclusions:

• CancerOmicsNet serves as a valuable AI platform for precision oncology.
• The system aids in developing novel therapeutic strategies for various cancer types.
• It facilitates the discovery of new approaches in precision oncology for diverse tumors and therapeutics.