2.1.2. Structure Based Virtual Screening (SBVS)

Drug candidates were retrieved from the PubChem database (https://pubchem.ncbi. nlm.nih.gov/) (accessed on 28 April 2022) to create the ligand library. There were 1200 structural analogues of Dabrafenib in the library. PyRx software was used to prepare the compounds library, which was converted to pdbqt format for virtual screening using Auto Dock Vina. The MMFF94 force field was used to minimize the energy of ligands. The

crystallographic structure of the targeted protein was retrieved from Protein Data Bank (https://www.rcsb.org/) (accessed on 1 May 2022) (PDB ID: 2WQN). After that, MGL tools were used to prepare macromolecule, which included removing Het atoms and water molecules, and the addition of polar hydrogen. The protein was examined for any missing residues. Furthermore, Kollman's charges were used to neutralize protein, and Gasteiger charges were calculated. Finally, for virtual screening of the compound library using Auto Dock Vina, a 1-angstrom grid box was built centered on the crystalline structure of protein at the point of co-crystal ligand (ADP) binding-site coordinates. The central xyz axis of the grid box was set to 80 × 80 × 80. Virtual screening was carried out after the targeted protein was prepared utilizing Auto Dock Vina's script-based technique. The exhaustiveness was set to 5 and the number of nodes was set to 20. The virtual screening was repeated twice to ensure the accuracy of docking results. In addition, docking protocol was validated by re-docking the co-crystal ligand with targeted protein. A RMSD value of less than 2 angstrom indicates the reliability of the docking pose. After completion of virtual screening, the output findings of the virtual screening module were analyzed and docking scores of drug candidates were compared to standard Dabrafenib. Only four compounds were found to have higher docking scores than standard Dabrafenib. The top hits were subjected to further analysis using deep learning algorithms. Deep learning models were used to predict drug affinity and determine the stability of protein–ligand complexes.
