Kinase Inhibitor Scaffold Hopping with Deep Learning Approaches

Area of Science:

• Medicinal Chemistry
• Drug Discovery
• Computational Chemistry

Background:

• Protein kinases are crucial drug targets, with many inhibitors targeting the conserved ATP-binding pocket.
• Scaffold hopping is a valuable strategy in drug design for generating novel chemical entities.
• The conserved nature of the kinase ATP-binding pocket offers opportunities for broad-spectrum inhibitor design.

Purpose of the Study:

• To introduce the SyntaLinker-Hybrid scheme for scaffold hopping in kinase inhibitor design.
• To leverage deep generative models for exploring novel scaffolds targeting the kinase hinge region.
• To generate kinase inhibitor-like molecules with diverse scaffolds while preserving key binding features.

Main Methods:

• Developed the SyntaLinker-Hybrid scheme for scaffold hopping.
• Utilized deep generative models to replace molecular fragments at the conserved kinase hinge region.
• Generated novel kinase inhibitor-like structures by hybridizing privileged fragments.

Main Results:

• Successfully generated kinase inhibitor-like molecules with novel scaffolds.
• Demonstrated the retention of binding features characteristic of existing kinase inhibitors.
• Validated the scheme's ability to explore diversified scaffolds across the kinase family.

Conclusions:

• The SyntaLinker-Hybrid scheme is effective for scaffold hopping in kinase inhibitor discovery.
• This approach facilitates the generation of novel chemical structures with potential therapeutic applications.
• The method can be applied to lead identification for various kinase targets.