A pharmacophore model is based on the structures of known ligands. Since there are no known ligands for this pocket, another approach is needed. Virtual screening is an option. If you do that, make sure you have an experimental plan for testing the selected ligands. Are you able to do real screening?

Good morning, Adam.

I'm pretty sure that we can do real testing (You mean that we will need to test the activity by in vitro assays, is this what you mean?). How do I should start if I want to use the virtual screening approach? I got the theoretical idea but, since this is my first time working on this field, I'm really excited and with a lot of doubts.

Thanks in advance.

Receptor based pharmacophore modelling is possible through LigandScout. In your case you dont knwo which residues are important for ligand binding so better to proceed with virtual screening. You can apply both methodologies to screen potential hits and finally select the common hits.

Good afternoon, Dear V. Kumar. 

Effectively, I don't know which residues are important for ligand binding, and also, these residues were identified by fpocket software, so, at least, I know which residues "could be" important (around 5 ~ 6). LigandScout allows me to use the residues in order to find potential ligands?

Thanks so much, Brayan.

Hi Antony,

Yes, I meant that you should follow up in silico screening with in vitro assays. I don't have the expertise to recommend virtual screening techniques and programs. Hopefully, someone else will be able to help with that. Make sure you limit your virtual screening to compounds you can actually get in solid form at a price you can afford, otherwise it will be a dead end.

Hello Antony,

So you actually know that it is an allosteric binding pocket. Have there been any mutational studies pointing out important amino acids in this area?

To generate a structure-based pharmacophore model ideally you need a pdb complex, where you are sure that ligand is positioned in certain area of the binding pocket and in a certain conformation. Then you can extract protein-ligand interactions from that complex.

In your case you don't have a complex. But you have a structure of the protein (or at least a model) and, on the other hand, QSAR and pharmacophore studies.

From ligand-based studies you can extract information important for ligands' activity, such as shape, location of H-bond donor, H-bond acceptors, etc

From your structure you can extract the shape of the pocket and its properties, whether it is lipophilic or hydrophilic, for instance, or which areas of the pocket would allow Van der Waals interactions and which H-bond networking.

Ligand-based and structure-based information combined together gives you a solid idea how the ligands sit in the pocket.

After that you could perform a docking study and prioritise those poses which explain both ligand-based and structure-based information. Docking study will provide you with a huge number of protein-ligand complexes from which you can extract structure-based pharmacophores. But be very careful and suspicious! You should only take a protein-ligand complex which makes sense! The one which explains available QSAR, pharmacophore and mutational studies!

Hope it helps!

Best,

Daria

If you know which sites are important, you can identify complementary pharmacophore features to develop models.  Old de novo design tools like Ludi can help with identifying the locations for the receptor binding features (there will potentially be 10s of locations per each h-bond interaction site at the receptor).  It is also possible to do this manually with pretty much any of the pharmacophore modeling software... but this manual development is rather difficult even for the people well versed with the art.  If you do not know which sites are important, then this also becomes a combinatorial problem.  You may have to develop 100s of pharmacophore models based on different subset of features etc.  I suggest you focus your initial resources for identifying which parts of the receptor active sites are critical, before getting into pharmacophore modeling.

If you do not have any information about the inhibitors then you can go for virtual screening protocols, and for your pharmacophore mapping you can use LigandScout 

As your starting material is the protein structure with no known ligand information. Perhaps you would like to look into the field of chemogenomics in which similar proteins interact with similar ligands. With that said, you can BLAST your protein sequence and see which other protein it is similar to and from there you can find from the literature or from bioactivity databases listed below for ligands that are known to interact with the similar protein. Another approach is to find proteins that share similar binding site and again find known ligands for those identified similar protein.

Bioactivity database:

http://www.bindinhdb.org

https://www.ebi.ac.uk/chembl/

Articles on chemogenomics:

http://www.nature.com/nrg/journal/v5/n4/abs/nrg1317.html

Articles on binding site comparison:

http://www.sciencedirect.com/science/article/pii/S0959440X1300208X

Dear guys, after a month of a really hard working-time in the lab, I know can understand what you were trying to explain me. I'm really embarrased that I woudn't be able to answer but I, truly, took into account what you mentioned. We finally picked up a pocket that was assessed and which had crystals and PDB info available, and FROM now, we are so eager to test what you are recommending.

Dear Daria, our current plan is based on what you said. We are going to combine QSAR (Ligand point of view) and protein info (Structure point of view) in order to further assess the phenomena of interaction among new candidates that we want to suggest and this allosteric pocket. 

Dear Osnan, 

Thanks for the suggestion. I'm not really used to Old de novo  design tools like Ludi, but I'm going to go beyond this and try to understand it. I hope I can message you as soon as possible in order to ask a little bit more about this, if you allow me. I need more training in pharmacophore modelling softwares.

Hey ya, Danish!

I was trying to install the LigandScout but I could'nt activate it. Do you have any idea of how I can use a trial version? I would like to prove it, at least, one time. Thanks for you recommendation.

Dear Chanin,

It's an eye-catching area, this that you were talking about. I think we can choose a chemogenomics approach too. I was thinking on using homologies as you mentioned but I would need to assess the homology % of the proteins. Do you know if there is a threshold that define which protein is useful for a chemogenomic approach? I mean, if after blast there is a ID homology % less than (ex.) 70%, it would be senseless using that protein or something like that?

Best regards!

And thank you all people!

Hi Antony,

For activating LigandScout you need to request the rial version, if you have institutional email id then you can easily request and get the license.

Regards