took DOCK6 to predict the docking structure. Following standard
DOCK6 protocol, the best docking conformation was generated as
1bma_dock.mol2. Then we employed Cyscore tools, and first add
hydrogen atoms to the protein PDB file using AddH as:$ ../bin/AddH 1bma_protein.pdb 1bma_protein_h.pdbNext, we submitted the protein and docking result to Cyscore
as:$ ../bin/Cyscore 1bma_protein_h.pdb 1bma_dock.mol2The output showed that Cyscore was1.7295. According to
the above introduction, Cyscore larger than3.5 implies that the
docking result may not be reliable. Then we computed the solvent
accessible surface grid to inspect the shape complementarity by:$ ../bin/CurvatureSurface 1bma_protein.pdb 1bma_protein_grid.
pdb1bma_protein_grid.pdb and the docking result 1bma_dock.
mol2 were visualized by Pymol. The coloring method for 1bma_-
protein_grid was chosen Color ->spectrum ->b-factor (Shown in
Fig.6). Thus we got the interface image. As shown in Fig.7a, there
were two regions in warm colors (Shown in the black circle), which
indicated they are concave surface and good for ligand binding.
The lower warm-colored region was partially occupied by a methyl
group, while the upper one had no contact with the ligand. Both
Cyscore and 3D visualization indicated the docking was unreliable.
The true binding structure was shown in Fig. 7b. Obviously
the real one is far from the docking structure. It shows highly
binding affinity (Cyscore¼5.5216), and occupies both warm-
colored regions.5 Notes
- If there are missing atoms in the input PDB file, will the
programs work?
Cyscore and CurvatureSurface will output results with
warning information in this case. Users should pay attention
to the warnings. If any of missing atom reported in the warn-
ings locates at the binding area, it may have impacts on the
prediction accuracy. AddH cannot work correctly in this case,
because it needs heavy atom’s coordinates to add hydrogen
atoms.
240 Yang Cao et al.