Table 1.
List of the discussed energy-based methods relying on geometry or a grid for target-focused pharmacophore modeling.
| Method | Cavity Definition | Approach | Clustering Method | Evaluation | Year | Refs. |
|---|---|---|---|---|---|---|
| Ph4Dock | cavity detection (Delaunay triangulation/α spheres) | electrostatic interactions (MMFF94 [26]) of charged dummy atoms | single-linkage | CCDC/Astex valida-tion set [27] d | 2004 | [18] |
| Pocket V2 | grid box around ligand (or user-defined pocket residues) | grid (Score) [28] | unclear clustering method a | CDK2, HIV1-PR, ER, 17b-HSD | 2006 | [19] |
| FLAP + BioGPS | grid box around ligand or FLAPsite detection | grid (GRID software) [17] | region-based energy minima | Patel set [29], DUD [30] d | 2007 | [23,31] |
| Tintori et al. | grid box around binding site | grid (GRID software) [17] | no clustering b (GRID minima + interpolation) | TrxR (MTB), HIV1 IN, HIV-1 RT dimer | 2008 | [21] |
| Hydro-Pharm | grid box around ligand (3 Å) | grid (ChemScore [25]) + MD-based hydration site feature reduction c | k-means | HIV1-PR, DHFR, FXa | 2012 | [22] |
| PharmDock | grid box around bound ligand (3 Å) | grid (ChemScore [25]) | k-means | PDB bind, DUD [30] d | 2014 | [24,32] |
| T2F-Pharm | grid box around ligand or user-defined center (& cavity point reduction) | grid (AutoDock) [33] | CNN [34] | Patel set [29] + A2A receptor | 2018 | This paper |
a Exact clustering method not specified in publication. b No clustering method used, points are reduced using the Minim and Filmap programs implemented in the GRID package, collecting all points within a certain energy threshold value and interpolating the closest ones. c Hydro-Pharms adds a second step to further restrict the pharmacophore points by calculating an overlap between the grid-based pharmacophores and molecular dynamics (MD)-derived hydration sites. d Evaluation focused on enrichment in docking/virtual screening, rather than evaluation of pharmacophoric features.