Proper docking protocols were presented for three known enzyme structures, human betasecretase (BACE1), Aromatase and pyruvate dehydogenase kinase (PDHK) using Autodock4.2 and Vina softwares. The validity of docking protocols was verified using a set of known active ligands and decoys for all three enzymes. Different energy minimization algorithms were performed prior to docking of each ligand in order to find out by which method a more reasonable correlation between binding energy and corresponding experimental activity of the compounds was obtained. The highest ROCAUC value was 0.916, 0.914 and 0. 833 when MM+-PM3 methods were applied as minimization method, whereas without minimization it was 0.127, 0.187, and 0.51 for PDHK, BACE-1 and aromatase, respectively. So a combination of molecular mechanics (MM+) and a semi-empirical method (PM3 or AM1) could promote the docking protocol in case of all targets. Protein ligand interaction studies using self-organizing map (SOM) were also conducted in order to reveal the validity of docking protocol and to evaluate its predictive ability in terms of distinguishing between ligands and decoys.