Binary Quantitative Structure-Activity Relationship Analysis to Increase the Predictive Ability of Structure-Based Virtual Screening Campaigns Targeting Cyclooxygenase-2

Abstract

Structure-Based Virtual Screening (SBVS) campaigns employing Protein-Ligand Interaction Fingerprints (PLIF) identification have served as a powerful strategy in fragments and ligands identification, both retro- and prospectively. Most of the SBVS campaigns employed PLIF by comparing them to a reference PLIF to calculate the Tanimoto-coefficient. Since the approach was reference dependent, it could lead to a very different discovery path if a different reference was used. In this article, references independent approach, i.e. decision trees construction using docking score and PLIF as the descriptors to increase the predictive ability of the SBVS campaigns in the identification of ligands for cyclooxygenase-2 is presented. The results showed that the binary Quantitative-Structure Activity Relationship (QSAR) analysis could significantly increase the predictive ability of the SBVS campaign. Moreover, the selected decision tree could also pinpoint the molecular determinants of the ligands binding to cyclooxygenase-2.