A machine learning-based KNIME workflow to predict VEGFR-2 inhibitors.

Area of Science:

• Medicinal Chemistry
• Computational Biology
• Drug Discovery

Background:

• Vascular Endothelial Growth Factors (VEGFs) and their receptors (VEGFRs) are crucial for angiogenesis, a process implicated in tumor growth.
• Overexpression of VEGFs/VEGFRs is common in cancers, making VEGFRs a target for anticancer drugs.
• Existing VEGFR inhibitors like sunitinib and sorafenib highlight the therapeutic potential of targeting this pathway.

Purpose of the Study:

• To develop and validate a computational model for identifying novel VEGFR inhibitors.
• To classify a large dataset of VEGFR inhibitors for drug discovery.
• To identify key structural features associated with VEGFR inhibitory activity.

Main Methods:

• Utilized the KNIME platform for descriptor calculation and machine learning model development.
• Employed classification algorithms including Linear Regression (LR), k-Nearest Neighbor (kNN), Decision Tree (DT), Random Forest (RF), and Gradient Boosted Trees (GBT).
• Evaluated model performance using accuracy, precision, recall, and F1 score on a diverse dataset from BindingDB.

Main Results:

• The classification models achieved high performance, with the best F1 scores reaching 0.87 for kNN, RF, and GBT.
• A dataset of 5120 VEGFR inhibitors was clustered into 10 subsets, revealing significant structural features.
• Identified key molecular fragments associated with both active and inactive VEGFR inhibitors.

Conclusions:

• The developed automated classifier is a valuable tool for screening and designing potential VEGFR inhibitors.
• This computational approach can accelerate the discovery of new anticancer agents targeting angiogenesis.
• The study provides insights into structure-activity relationships for VEGFR inhibitors.