Efficient Virtual Screening Using Multiple Protein Conformations Described as Negative Images of the Ligand-Binding Site

Abstract

The protein structure-based virtual screening is typically accomplished using a molecular docking procedure. However, docking is a fairly slow process that is limited by the available scoring functions that cannot reliably distinguish between active and inactive ligands. In contrast, the ligand-based screening methods that are based on shape similarity identify the active ligands with high accuracy. Here, we show that the usage of negative images of the ligand-binding site, together with shape comparison tools, which are typically used in ligand-based virtual screening, improve the discrimination of active molecules from inactives. In contrast to ligand-based shape comparison, the negative image of the binding site allows identification of compounds whose shape complements the shape of the ligand-binding cavity as closely as possible. Furthermore, the use of several target protein conformations allows the identification of active ligands whose shape is not optimal for crystallized protein conformation. Accordingly, the presented virtual screening method improves the identification of novel lead molecules by concentrating on the optimally shaped molecules for the flexible ligand binding site.

Ultrafast protein structure-based virtual screening with Panther

“…In this work, the same protein targets were used as in the preceding work [5,22] in order to be able to compare the performance of the developed algorithm to the previously used approaches. The 3D structures given in Directory of Useful Decoys [26] (DUD) and the Directory of Useful Decoys: Enhanced [27] (DUD-E) for androgen receptor (AR), estrogen receptor alpha (ERa), glucocorticoid receptor (GR), mineralocorticoid receptor (MR), peroxisome proliferator activated receptor gamma (PPARc), progesterone receptor (PR), retinoid X receptor alpha (RXRa), cyclo-oxygenase 2 (COX2), and phosphodiesterase type 5 (PDE5) were downloaded from the Protein Data Bank [28] (PDB) ( Table 1).…”

“…Shape and electrostatic comparisons of the Panther models to DUD and DUD-E molecule sets were performed with ShaEP [12] as described previously [5,22]. The similarity searches were done utilizing equally the shape and the electrostatics, this is, as the early enrichment tends to be highest in most of the cases when the shape and the electrostatics are contributing equally to the binding [22].…”

“…When the performance of the novel Panther method was compared with the NIB (negative image-based) method [5,22] (our previous approach), the AUC values were either equal or marginally better (Table 2). With the ERa antagonists, PR and COX2, the AUCs were slightly higher.…”

“…Unfortunately, rapidity and accuracy often do not coincide. Various studies have suggested that scoring functions do not always separate active ligands from inactive molecules which may lead to docking results that may vary from target to target [1][2][3][4][5][6].…”

“…In our previous studies, we showed that our simple atomistic space-filling model was effective [5,22]. The earlier creation of a negative image of the ligand-binging area (the model) relied on VOIDOO/FLOOD [25], which was initially designed for cavity searching, cavity volume analysis, and filling found cavities with water molecules mainly for crystallographic purposes.…”