Discovery of novel acridine based inhibitors of Bruton's tyrosine kinase (BTK) via pharmacophore modeling and machine learning-driven virtual screening followed by molecular dynamic studies.

Bruton's Tyrosine Kinase (BTK) has been recently recognized as an important drug design target for treating B-cell malignancies. Unfortunately, drug resistance is making the treatment of B-cell malignancies challenging. In this study, we employed a dual machine learning drug design approach that involves two parallel branches: the QSAR-GFA modeling (Branch 1) which engages the development of an interpretable structure activity relationships, and the KNIME® based machine learning modeling (Branch 2) which aimed at achieving high predictive accuracy. Subsequently, six machine learning modules, encompassing; the Random Forest model, the XGBoost Model (XGBoost), the Naive Bayes model, the k-Nearest Neighbors model, and the Support Vector Machines model were applied. Later, experimental validation through cytotoxicity assays against Raji lymphoma and K562 leukemia cell lines revealed several compounds that are exhibiting notable biological activity. For instance, compounds 166 and 168 which are acridine based displayed IC