A novel real-time editor for protein-ligand binding affinity prediction using structural comparison

Identifier:  TFM:2340

Handle: https://hdl.handle.net/20.500.11797/TFM2340

Authors:  Parade Patil, Kuldeep Shivaji

Abstract:
This thesis presents a novel method for protein-ligand binding affinity prediction, integrating structural comparisons using Graph Edit Distance (GED) and molecular descriptors. A real-time graphical editor is developed for visualizing molecular structures, calculating GED, and comparing binding affinities. Experimental results on the SARS-CoV-2 main protease dataset demonstrate the approach’s efficiency, combining interpretability and prediction accuracy. Additionally, a K-Nearest Neighbors (KNN) model is utilized for affinity prediction, supported by quantitative error analysis and visualization tools. This research aligns with ongoing efforts at Universitat Rovira i Virgili (URV) to innovate computational drug discovery methods, emphasizing transparency and practical applications in cheminformatics.