Figure 5.

Different ways of associating targets with DrugBank drugs. The site-based association uses binding site similarity as a metric for assigning function to a protein. Consensus pocket prediction exercise yields two pockets for the protein fadD19 (Rv3515c), 1 pocket for cmtR (Rv1994c), and two pockets for ald (Rv2780). Site matching exercise on PDB site library gave 43 ligands that could bind with fadD19. Ligand hits were then pruned by establishing chemical fingerprints on DrugBank molecules. Cobimetinib and darifenacin came out as the best repurposable drug candidate for fadD19. No hits were identified for ald and cmtR. Fold based annotation involves mapping drugs to the targets from the known drug binding proteins. The 3D structure of 4,234 proteins was built using MUSTER and MODELER, of which one hit, very long-chain acyl-CoA synthetase, binding to bempedoic acid share a fold level similarity with fadD19. No hits were identified for ald and cmtR. Finally, sequence derived analysis uses the sequence search option in ChEMBL to identify bioactive compounds. 10 drugs were identified for fadD19, 3 for cmtR. No hit was obtained for ald.