Identi fication of Pockets on Protein Surface to Predict Protein–Ligand Binding Sites - Identification of Ligand Binding Site and Protein-Protein Interaction Area - page 28

Input

Given protein structure

Project structure into a 3D grid

Classify grid points to “inside protein”,
“near surface” and “solvent”

Scan 7 directions for each solvent grid
point; calculate the number of SSS event

N SSS >= 5

Yes

The grid point is located in a pocket

Cluster such grid points according to
spatial similarity

Rank the clusters (pocket sites) by
their sizes

LIGSITE cs

Identify potential ligand-binding residues
(Residue mapping)

Re-rank the pocket sites by the
conservation score of around residues

LIGSITE csc

Output

Identified pocket sites
Potential ligand binding residues for each pocket site

Fig. 2.2 The detailed work fl ow of LIGSITE cs/c . The identified pocket sites in LIGSITE cs are
ranked by the pocket size and then re-ranked by residue conservation scores in LIGSITE csc

to 9 (more conserved). This ConSurf-HSSP database pre-calculates the conservation
score for all the PDB files in the PDB. However, if the users submit a new protein
structure without any PDB ID, it is impossible to retrieve the conservation score
from ConSurf-HSSP and thus the last re-ranking step LIGSITE csc could not be
applied. In such cases, geometric ranking LIGSITE cs will be applied and the pocket
sites are thus ranked by the pocket sizes rather than conservation score in the end.
The whole process of LIGSITE csc is illustrated in Fig. 2.2 in details.