Predicting Kinase Inhibitor Resistance: Physics-Based and Data-Driven Approaches.

Area of Science:

• Computational biology
• Drug discovery
• Molecular modeling

Background:

• Drug resistance is a major challenge in treating cancer, viruses, and bacteria.
• Protein mutations that affect drug binding are a common cause of resistance.
• Anticipating these mutations can aid drug development and clinical strategies.

Purpose of the Study:

• To evaluate computational methods for predicting changes in ligand binding affinity due to protein mutations.
• To assess the accuracy of these methods in identifying drug-resistant mutations for the Abl kinase target.

Main Methods:

• Utilized three distinct structure-based computational approaches: first-principle statistical mechanics, mixed physics- and knowledge-based potentials, and machine learning.
• Applied these methods to predict binding affinity changes upon mutation for the Abl kinase.

Main Results:

• The computational methods accurately estimated binding affinity changes caused by mutations.
• These approaches successfully identified mutations conferring resistance to Abl kinase inhibitors.
• Remarkable accuracy was observed in predicting resistance-causing mutations.

Conclusions:

• Computational methods can reliably predict drug resistance mutations.
• These complementary approaches offer a powerful tool for anticipating resistance in various target proteins.
• Routine prediction of resistance mutations will benefit drug development and patient treatment.