kinCSM-RTK: Machine Learning-Based Screening of Receptor Tyrosine Kinase Inhibitors in Drug Discovery

Area of Science:

• Biochemistry
• Computational Biology
• Pharmacology

Background:

• Receptor tyrosine kinases (RTKs) regulate crucial cellular functions like differentiation, migration, and proliferation.
• Dysregulated RTK activity is implicated in diseases such as cancer and neurological disorders.
• Small molecule inhibitors are vital therapeutics for managing conditions with overactive RTKs.

Purpose of the Study:

• To address the challenges of high cost and time in preclinical drug development for RTK inhibitors.
• To develop a computational tool for predicting the inhibitory potency of small molecules against RTKs.
• To facilitate the efficient shortlisting of potential drug candidates for RTK-targeted therapies.

Main Methods:

• Developed kinCSM-RTK, integrating two machine learning models.
• Models predict pKi and pIC50 values, quantifying small molecule inhibitory potency against RTKs.
• Validated model performance using 10-fold cross-validation and an independent blind test.

Main Results:

• Achieved high predictive performance for pKi (Pearson's r = 0.773 for 10-fold CV, 0.762 for blind test).
• Demonstrated robust pIC50 prediction (Pearson's r = 0.773 for 10-fold CV, 0.768 for blind test).
• Post hoc analyses revealed aromatic interactions correlate with stronger RTK inhibition.

Conclusions:

• kinCSM-RTK offers a robust and generalizable approach for predicting RTK inhibitor potency.
• The findings provide insights into structure-activity relationships for RTK drug design.
• The models are publicly accessible via a web server for broader research use.