Drug Discovery for Thyroid Cancer Using CNN and GNINA

Abstract

Thyroid cancer often caused by the BRAF V600E mutation, which activates the MAPK/ERK signaling pathway that promotes tumor growth. Targeting this mutation using kinase inhibitors such as Vemurafenib, Dabrafenib, and Sorafenib has shown effectiveness. Traditional docking tools like AutoDock Vina have limitations in accurately predicting binding affinities. To overcome this, the project uses GNINA, a deep learning-based molecular docking framework that applies Convolutional Neural Network (CNN) scoring functions to improve protein–ligand interaction predictions. Through redocking, cross-docking, and whole docking approaches, to identify the most effective treatments, we evaluated how well different inhibitors bind to the BRAF V600E protein. The study shows that by blending deep learning with molecular docking, researchers can fast-track the discovery of new, more effective therapies for thyroid cancer.