Studies and improvement of molecular docking methods in rigid and flexible receptors

  • Armelle Le Gall

Student thesis: Doctoral Thesis


Computational chemistry is a field of chemistry that uses computers to model chemical structures and properties and as such has been used in drug discovery since the eighties.1 Computational chemistry within the field of Drug Design can be sub-divided into a number of different areas, including target identification, high-throughput screening analysis, de novo design, molecular docking, activity predictions, etc. These methods aim to predict the properties of molecular systems, such as the energy, the activity profile or the physico-chemical properties. The main objective of computational chemistry in the pharmaceutical industry context is to help decision making by guiding research scientists as to which molecules to synthesise or test.2 In the case of structure-based drug design, computational chemistry relies on structural biology to provide structural data, for instance crystal structures of protein-ligand complexes. These crystal structures give information about the interaction of the ligand and the protein binding site and can be used in the design of subsequent and more optimised molecules. The use of rigid receptor molecular docking is exemplified in this work through the investigation of the LTA4H system. A number of analogue compounds of 5 were docked in LTA4H successfully and it was shown that molecular docking could be used as a tool for stereochemistry assignment for the closely related LTA4H ligands 10a-d. In this work, molecular docking predicted the isomers associated with measured potencies and these predictions were confirmed by subsequent experimental data. One drawback of rigid receptor docking is that it does not account for the flexibility of the target. The investigation of protein flexibility in molecular docking, therefore, followed. Standard protocols from the Schrödinger modelling suite were investigated first and the results were benchmarked for the fXa and CDK2 dockings, which were compared to the published Schrödinger results. The Induced Fit Docking (IFD) protocol was tested with default settings, and with automated and manual truncations of amino acids. These experiments concluded that the selection of amino acids could be improved as was also the case for the scoring function, IFDScore. Therefore, in this work a new approach for the automatic selection of amino acids for flexible exploration of the binding site was introduced. In addition, a modified scoring function was investigated and applied to fXa, CDK2 and UPPS systems. The new selection and scoring protocols showed some advantages over the standard Schrödinger IFD protocol.
Date of Award10 May 2017
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde

Cite this