Define at least two pharmacophore constraints: (1) search space and (2) key pivoting interaction. If there are multiple possibilities, explore them all. In the absence of relevant experimental data, assume that compounds belonging to the same chemotype bind preferentially to the same binding site.

You can try Fpocket, it is able to perform simple pocket detection. I have never tried it but I use similar software LigandScout Advance version to predict the pocket by function Apo-site prediction. Other ways are you can try to do the literature review related to your protein or try to find the information related to the pocket in Uniprot (a database for protein sequences and functions).

https://bioserv.rpbs.univ-paris-diderot.fr/services/fpocket/

http://www.inteligand.com/ligandscout/

https://www.uniprot.org/

Firstly, search literature to find the binding site of the corresponding protein.

Otherwise, you can identify protein pockets through these in silico tools:

http://www.vls3d.com/index.php/links/bioinformatics/binding-pockets

Thank you so much everyone.

This's an interesting question. There are several ways to find the binding site of proteins.

1- As the respected colleagues answered, using online databases, you can do it.

2- For all crystallographic structures available in the PDB database, the binding site of target proteins can be accessed through the paper that published the relevant protein. In this case, you can search for your protein in PDB to find its binding site.

3- Using blind docking. Blind docking is a procedure in which you can dock several ligands with different backbones into target receptors to evaluable the possible interaction between these ligands and protein surfaces in different areas. Based on docking energies, you can categorize your ligands and find relevant sites that are the most important locations for interactions of your ligands. This method, in some cases, works fine, and it is a reliable way to find the binding site of proteins.

4- Using the PDBbind database. PDBbind database consists of many experimentally assayed binding modes of ligands and receptors. You can download it and check its content for relevant data.

5- Using text-mining approaches. This method will enable you to use artificial neural networks to excavate the literature to find the binding mode of ligands into your target receptors. In this way, you can find many papers dealing with your project and use their content to predict the binding site of your protein.

6- Using PSI-BLAST. This method is a time-consuming procedure but returns validated results. For this case, you can blast the sequence of your query against the PDB database and find more similar proteins. Based on the location of other protein active sites that showed similarity with your protein sequence, you can find some binding sites in your protein structure. Please note that the alignment output can be submitted to ENDsprit software to compare the results and align the studied sequences' binding site.

Helpful papers:

These papers can help you to know more about the prediction of protein active sites:

1- Article Protein active site prediction for early drug discovery and designing

2- Article Exploring the Computational Methods for Protein- Ligand Bind...

3- Article Advances in protein structure prediction and design

4- Article Predicting Protein Ligand Binding Sites by Combining Evoluti...

Good Luck

Rasouli. H