AutoSite is a computational method and tool that is used for the identification of binding sites of small molecules in target proteins [1]. It comes with the MGLTools2 package along with ADFR and AGFR. In this article, we will predict the binding sites in a receptor using the Autosite tool.
The potential binding sites in the receptor are computed and ranked as:
(No. of fill points X buriedness^2 )/(Radius of gyration)
We are using human serum albumin in this tutorial for binding site prediction.
Preparing receptor
- Download the structure from PDB (here, 2BXA).
- Remove extra chains and hetatoms.
- Prepare the PDBQT file of the receptor following the same procedure as used in the preparation of the receptor for docking. You can find a detailed procedure in this article. As you don’t need any grid file for this tutorial. Therefore, you can skip the grid box defining part for now.
Executing AutoSite
Open a terminal (Ctrl+Alt+T), change to the directory where you have saved your receptor PDBQT file, and type the following commands:
$ cd Downloads/
$ /home/user/Downloads/mgltools2_x86_64Linux2_1.0/bin/pythonsh /home/user/Downloads/mgltools2_x86_64Linux2_1.0/MGLToolsPckgs/AutoSite/bin/AS.py -r 2BXA.pdbqt
Don’t forget to replace the ‘user’ in the above command with your username.
As a result, it will output rank-based receptor files named as ‘_cl_<rank>.pdb‘ for the predicted pockets and ‘_fp_<rank>.pdb‘ for the predicted ligand.
References
- Ravindranath, P. A., & Sanner, M. F. (2016). AutoSite: an automated approach for pseudo-ligands prediction—from ligand-binding sites identification to predicting key ligand atoms. Bioinformatics, 32(20), 3142-3149.
