Artificial Intelligence Predicts Sensitivity of EGFR Novel Mutations to Tyrosine Kinase Inhibitors

*Correspondence author: (e-mail: [e-mail]).

Abstract

Background: Aberrant EGFR kinase activity, resulting from somatic mutations, is associated with driver phenotype, making EGFR a critical target in cancer therapy. This study aims to predict the response of tyrosine kinase inhibitors (TKIs) in cancer patients through a combination of molecular dynamics (MD) simulations and machine learning (ML) techniques.

Material and Methods: We developed automated and scalable pipeline for homology modelling, molecular docking, and MD simulations to elucidate the interactions of EGFR mutants (N = 473) with TKIs. We generated features from the in-silico analysis to feed into ML algorithms resulting into a TKI sensitivity prediction model with high accuracy and sensitivity.

Results: Our analysis of about 500 EGFR mutation models, molecular docking scores, and about 24 micro-seconds MD simulation reveals interesting correlation between in-silico observations and clinical TKI sensitivity. We were able to derive an artificial intelligence model with high accuracy for prediction of EGFR TKI sensitivity.

Conclusion: The structural features observed in EGFR mutant complexed with TKIs offer valuable information for the AI-based prediction of drug sensitivity against novel and rare EGFR mutants. Further refinement and validation of this model may provide valuable solution to predetermine the drug sensitivity of patients in clinics.

No conflict of interest has been declared by the author(s).

Publication History

Article published online:
08 July 2024