Method for Galaxy7TM

Galaxy7TM performs flexible GPCR-ligand docking by structure refinement. With a given
model structure of GPCR and a ligand structure, 10 refined GPCR-ligand complex structures
are generated. After initial docking by GalaxyDock, the resulting complex structures are
further refined by a method based on GalaxyRefine. Both the receptor and ligand are treated
flexible in this refinement process which is based on protein sidechain perturbation and
overall structural relaxation. Final models are selected based on ranks by two energy functions,
one used for structure refinement and the other used for ligand docking.

Inputs for Galaxy7TM

User Information

First, you have to enter the following information required for modeling.
  • Job name: Enter job name for your recognition.
  • E-mail address: GalaxyWEB server will send progress reports of your job.


You have to provide a GPCR structure in PDB File Format.
Input GPCR structure with up to five gaps are allowed if a full sequence file in FASTA format
matching the PDB residue numbering is submitted together.
The server builds the gaps from the sequence by a simple method. Long gaps (> 3 residues) are
recommended to be built using other programs (e.g. GalaxyLoop) before submission because
the initial structure can affect the docking performance.
Sequence file is not required if input GPCR structure does not have any gap.
Any HETATM lines in the PDB file are neglected.

An input 3D ligand structure can be submitted in MOL2, PDB or XYZ format.
The server converts a ligand input in PDB or XYZ format to MOL2 using Open Babel.
Ligand file containing only one structure is accepted.

Option for assigning binding pocket residues

If binding pocket residues are known or can be predicted, up to 10 residues can optionally be submitted.
The residue numbers must match those in the input PDB file.
If you need to obtain generic residue numbers from the residue numbers in your PDB file, please use the
GPCRdb PDB file residue numbering tool.
(You can use another tool to get generic residue numbers from your sequence.)
These residues will be used for locating the docking box. The docking box, a cube of 22.5 angstrom
is centered at the average position of the input residues. If binding pocket residues are not submitted,
they are predicted based on the GPCR structure in the PDB closest to the input GPCR structure.

Outputs for Galaxy7TM

Predicted Structures

The 10 selected GPCR-ligand complex models are shown. The input GPCR structure is shown in white
as a reference. The final models are in green with each bound ligand in black. You can also download
the PDB files.

Model Information

Refinement energy and docking energy that are used to generate and select the models are provided.
The 10 conformations are selected by the sum of rank in refinement energy and half of the rank in docking energy.
This score was trained to predict the docking accuracy, and the models are ordered by this score. In addition
to this order, it is recommended to focus on ligand conformations that share common interactions with others.
Ligand RMSD of each model from Model 1 is also reported in the table.
A link to a schematic diagram of GPCR-ligand interactions drawn by using LIGPLOT is provided for each model.
A link to a list of protein residues making close contacts (< 4 angstroms from the ligand atoms) is also provided.


Users can download 10 structures generated by Galaxy7TM.

References for Galaxy7TM
  • G. R. Lee and C. Seok*, Galaxy7TM: flexible GPCR-ligand docking by structure refinement, Nucleic Acids Res. 44 (W1), W502-506 (2016). [LINK]

Related servers
  • Hybrid MM/CG webserver (for the setup and running of short multiscale molecular dynamics simulation for model refinement) [LINK]