Support Vector Machines in Virtual Screening ...

Support Vector Machines in Virtual Screening for Therapeutic Exploration Using Radial Basis Function (RBF) Kernel for Kinase Inhibitor Discovery

Author Name : Madan Mohan Tito Ayyalasomayajula, Sailaja Ayyalasomayajula

ABSTRACT Virtual screening (VS) is a fundamental technique in modern drug discovery, essential for identifying potential therapeutic molecules from extensive compound libraries. Among the various computational methods, Support Vector Machines (SVMs) have proven to be a robust machine-learning approach for VS tasks due to their capability to classify high-dimensional data effectively. This review centers on using SVMs with the Radial Basis Function (RBF) kernel specifically for virtual screening applications, focusing on discovering kinase inhibitors. We delve into the benefits and drawbacks of SVM-RBF models, examining their role in kinase inhibitor discovery and offering practical insights into implementing SVM within virtual screening workflows. Additionally, we compare SVM with other machine learning algorithms to assess its effectiveness in kinase inhibitor exploration. The discussion extends to the significance of feature selection and molecular descriptors in refining the predictive accuracy of SVM models. We address critical aspects such as kernel parameter selection and strategies for managing imbalanced datasets, which are pivotal for optimizing model performance