I understand that for pharmacophore modeling to be carried out, three or more known inhibitors or reported compounds with high activity against the target protein should be known. But in a case whereby the target protein is novel and only one inhibitor of the target has been reported. Can pharmacophore modeling be carried out after a virtual screening to understand the features of the top hits?
If I understand your question correctly, you are going to perform a virtual screening using some other computational methods (e.g. docking) and from generated hits you will perform pharmacophore. Idea is not bad, something is better than nothing! I would say, choose your cut-off for VS very carefully (if you are going to perform docking based VS, first dock known ligand, estimate binding affinity and based on that set cut-off). Remember, for pharmacophore, inactive hits (molecules) is also important, so you should select some inactive hits as well. I am not sure how much robust your model will be, but at least you will get some ideas about which kind of ligands will work for your protein.
Riaz A. Khan Kazi Amirul Hossain Martiniano Bello Thank you for your comments
I am sorry to differ but the very premise of pharmacophore modelling is to utilize a set of known binders with known, experimentally derived IC50 values as a training set and derive the pharamcophore model. Hence, this technique is used usually when the receptor structure is not known but a number of known binders (natural substrates, analogues, ligands etc.) are known. Structure based pharmacophore modeling is also possible when a number of cocrystallised ligand-protein structures are available. However, using the results of virtual screening, which is in itself a predictive exercise, for generating another predictive model (pharmacophore) is not scientifically sound. The pharmacophore so generated would not be based on experimentally proven training set with known biological activity and hence will be meaningless.
- Badges
- Science topic