Scoring Functions in Docking Experiments

Introduction

Molecular docking is a computational technique used to predict the binding orientation or native pose of a molecule while interacting with another molecule to form a stable complex. This technique is routinely carried out to study various kinds of inter-molecular interactions between ligand and protein, ligand and DNA, protein and protein, protein and DNA. In the pharmaceutical field, it is often used to find out the binding orientation of a lead molecule (ligand) to its target protein so as to understand the mechanism of action as well as to predict the binding affinity that plays a vital role in the rational drug designing (RDD). Each docking algorithm comprises one or more ‘scoring function(s)’, which plays the central role in exploring the correct binding orientation and understanding the outcome of the docking experiment. In docking, scoring function can be defined as a mathematical function that is used to compute a representative score and to decide a binding pose, by evaluating the favorable (rewarding term) and unfavorable (penalty term) inter-molecular interactions found in a docked complex. The typical components of scoring functions are hydrogen bonding (illustrated in Figure 1), electrostatic interaction, van der Waals interaction (Figure 2), desolvation effect (Figure 3) and loss of torsional entropy upon binding (Figure 4).

Figure 1.

Hydrogen bonding

Snapshot taken in Discovery Studio software (DiscoveryStudio, 2007) for demonstration purpose only.

Figure 2.

van der Waal interactions

Snapshot taken in Discovery Studio software for demonstration purpose only.

Figure 3.

Desolvation effect

The figure was built using the snapshots taken in Discovery Studio software for demonstration purpose only.

Figure 4.

Loss of torsional entropy upon binding

The figure was built using a snapshot taken in Discovery Studio software for demonstration purpose only.

Background

The basic aim of a scoring function is to find a docking score that should somehow represent the experimentally-determined binding affinity of the docked ligand.

The scoring functions are intended to accomplish the following mentioned tasks:

• 1.

  To predict the absolute binding affinity between a ligand and a protein or at least the relative binding affinities among the multiple ligands and a protein.

• 2.

  To decide the correct binding orientation (native pose) and position of a ligand in the active site of the protein.

• 3.

  To identify the potential lead compounds against a target protein by screening a huge ligand database.

The desired properties of a scoring function are as follows:

• 1.

  Ideally, the computed docking score should be significantly correlated with the experimental binding affinity of the ligand.

• 2.

  It should precisely distinguish between correctly and incorrectly docked structures.

• 3.

  The calculation of scoring function should be sufficiently fast, so that it can be used smoothly to screen huge databases (Huang, Grinter, & Zou, 2010).