From the literature you will have to find the ligand or ligand(s). If that particular molecule is commercialized than to further improve its stability and binding affinity or its effect a compound library can be built based on structural derivatives of that drug or ligand, and further analysis could be done. Also if in your protein structure there is a ligand already bound  to stabilize the macromolecule or to give some effect than that itself could be used as a base molecule to create a library for virtual screening.

I am facing similar problem. In the absence of ligand, you can try using PyRx for virtual screening. You can use PyRx output as a ligand in other docking platform.

You will still need to use other platform though. Hope it will be helpful

@Murli

I'm not familiar with Pyrx ... I tried to install it but it demands ligand and macromolecule.

could you please elaborate how to use macromolecule only?

Thanks in advance

Dear Ahmad,

If you have a target protein that is druggable, you might want to perform high-throughput docking (you could also call this virtual screening or structure-based drug discovery) against a large chemical library as to identify hit compounds. Once you have obtained some hits the next step would be hit-to-lead transformation where they are further subjected to lead optimization via bioisosteric replacements in which various substituents are changed to different functional moiety (i.e. methyl, ethyl, propyl, nitrile, hydroxyl, methoxy, etc.) as to optimize their ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) properties in addition to their bioactivity against the target protein. A great review of the drug design process is available in the review article below:

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3058157/

At the onset, preparing a chemical library maybe a daunting task if you are starting out. Some chemical library that you could use right away for your virtual screening could be:

1) NCI diversity sets (each set contains about ~1500 non-redundant structures derived from an initial set of ~100,000 structures)

https://wiki.nci.nih.gov/display/NCIDTPdata/Compound+Sets

2) BindingDB (a chemogenomic resource that contains bioactivity data for proteins and their respective ligands)

http://www.bindingdb.org

3) ZINC (contains 35 million compounds that are ready for docking and are all purchaseable)

http://zinc.docking.org

4) Other compound sets can be found from the link below:

http://htsc.wustl.edu/collections.html

In performing Docking, I also agree with Murli on using PyRx as it automates the docking process and are user-friendly if you are starting out. The conventional approach would be to manually perform the ligand and protein pre-processings of adding Gasteiger charges to the former and Kollman charges to the latter but PyRx does this automatically upon importing them. Before installing PyRx you might want to install some dependencies such as: Microsoft Visual C++ 2008 Redistributable Package, Python, MGL Tools, etc. Please see the forum for extensive information: 

http://mgl.scripps.edu/forum/viewforum.php?f=25

When you had successfully install it, you can proceed with protein-ligand docking which requires both ligands and protein. I also would recommend that you go through the Getting started video from Sarkis Dallakian (the developer of PyRx) and the other videos that he prepared before venturing into your own endeavor:

https://www.youtube.com/user/dallakyan/videos

There's a couple of options. The first option is just to dock a set of ligands from a library using AutoDock or another tool setting the grid to center on your perspective binding site. The problem is that docking is slow so and therefore you can only  scan a fairly narrow library of compounds . If you already knew an active compound, you could use ligand based similarity screening to reduce the size of the library prior to docking. Since you don't know what an active looks like for your protein, the computational cost of docking is going to be a problem in virtual screening.

There is another option, which is to use shape based ligand searching against the binding pocket. A negative image of the binding pocket is generated and conformers of the ligands are matched against the image. This is the approach taken in SLIM and BSP-SLIM (link below). This approach is more than two orders of magnitude faster than docking and actually more accurate. A bound structure or active ligands should not be needed. The problem is that it uses proprietary OpenEye software for which the license is no longer available. If alternative is found to get it working again I will let you know. 

http://onlinelibrary.wiley.com/doi/10.1002/prot.23165/full

I think Chanin gave quite a great description that can be referred by beginners

IN PyRx

you can use your protein as macromolecule

You can use multiple compounds to be screened as ligands. Try putting a small set of small molecules like one provided by NCI in a single directory (folder). Select this directory as ligand.

Screen using PyRx and then look for compounds that bind to the binding pocket of your target.

You can then use these hits for improving your screenin by using hits as ligand.

@ all

Thank you for the marvelous help.

@ Jeffrey

I found this online server for BSP-SLIM, I'll start with a large compound set in BSP-SLIM then use the result as a docking library.

http://zhanglab.ccmb.med.umich.edu/BSP-SLIM/

The online server for BSP-SLIM won't work until a workaround is made for the Openeye software, which is no longer available.. I do not know when that will be. However, once a workaround has been found I will let you know (I am currently in the Zhang lab). I only mentioned it because it seems like an ideal tool for these cases (pocket known but no actives available)

As an alternative, you can try docking with the NCI diversity sets suggested by Chanin.

Dear Ahmad,

Some databases of protein ligands (not only mentioned by Chanin) are linked here:

http://www.uwm.edu.pl/metachemibio/index.php/protein-ligands

in the absence of known ligand one can use the server http://www.scfbio-iitd.res.in/software/drugdesign/raspd2.jsp (method-B) for virtual secreening and preparing ligand library..

server doesn't work

The license issue with BSP-SLIM has been fixed. Openeye software is free for academic use but the wrong license was installed.

Ahmad, can you send me the file so I can try it out on the NCIBI diversity set and we can eventually set the BSP-SLIM server to allow a virtual screening option (currently only runs with a single ligand)?

Use  pubchem database, drug bank, marine databases which are freely available as small molecule source

You might want to check out my lab's free Gypsum-DL software: http://durrantlab.com/gypsum-dl/

If you've got a list of molecules in SMILES or flat SDF format, Gypsum-DL will convert it to 3D models ready for docking. The program also considers alternate protonation, tautomeric, or chiral forms, which can be helpful in some cases. Best of luck.