Fig 1.
Schematic illustration of the WaterDock 2.0 bridging water prediction pipeline.
Table 1.
The H-bond saturation limit enforced on the various functional groups implemented in the WaterDock 2.0 pipeline.
Table 2.
The dataset used to validate the ligand-directed Waterdock 2.0 algorithm.
Fig 2.
The MD (A-E) and modelled (F-J) hydration structure of five polar functional groups. (A, F) carbonyl, (B, G) carboxyl, (C, H) ether, (D, I) phosphoryl, (E, J) imine.
Fig 3.
The location of water-sites around different amine groups as produced in MD.
Fig 4.
(A) The distribution of hydration sites around the hydroxyl functional group from MD simulation. (B) The modelled hydration structure of the hydroxyl functional group.
Fig 5.
Scatter plot showing the water-site’s Vina docking score against the distance to the nearest crystallographic water on application of the WaterDock 2.0 pipeline to the Astex Diverse Set.
The 2.0 Å distance cut-off is plotted as a vertical dotted line and the Vina cut-off score of -0.55 kcal/mol is plotted as a horizontal dotted line. The lower left quadrant thus signifies crystallographic waters molecules correctly identified within the training set.
Fig 6.
The location of the crystallographic waters from two structures of PIM-1 with PDB accession codes 1XWS and 2BIK, where the crystallographic waters from the two structures are shown in red/blue and the predicted waters from the two runs are shown in green/yellow, respectively.
A false-positive result arises from two sites (green) predicted adjacent to the same crystallographic water (red/blue).
Table 3.
The results of the validation dataset of eleven protein holo-structures.
Table 4.
The results of the OppA dataset of structures used to allow comparison of new prediction protocol to AcquaAlta and the original WaterDock methodologies.
Fig 7.
Snapshots of the PyMOL plugins developed for WaterDock pipelines.