In docking we are trying to obtain the binding conformation of a molecule against a target. That is why when validating a docking algorithm you always compare the results of the algorithm against experimentally determined protein-ligand complexes. Your algorithm should be obtaining something similar to the experimentally determined structure.

In my opinion it doesn't make sense to try to use protein structures without a cocrystalized ligand to validate a docking algorithm. What are comparing then? In the next question you talk about other cavities, are you only interested in finding binding cavities? That is a different matter, and only related to docking if in your algorithm you don't need to specify the binding pocket and the algorithm tries to find it for you.

Hi Rajeev

RMSD is a relative term and can be only used when we have a co-crystallized ligand. One cannot extrapolate the concept of RMSD for the validation of a docking algorithm in case of apo-protein.Regarding your second question I'm not pretty sure what exactly you are enquiring for?

But there must be some parameters for validation in case of apoproteins or when we try to dock ligands in different cavities other than that of co-crystallized ligand. 

The ultimate validation of any computational method is experimental validation. If your algorithm proposes that a ligand binds into a novel pocket and you can obtain an X-ray or NMR structure that reproduces such binding then you struck gold. However traditional docking is basically just a search for the specific conformation of a compound when it binds into a protein pocket. As Mohsin Yousuf Lone mentioned, we use RMSD to compare to a different conformation, normally an experimentally obtained binding conformation in order to prove the correctness of the docking procedure. In those cases you mention in the comment, you simply have no reference to compare to and so traditional validation techniques make no sense. I know of no computational validation technique applicable to the cases you mention.

If you have developed a new docking algorithm then you'd need to evaluate its performance and accuracy using a representative, that is, large and diverse, set of protein-ligand complexes. This would implicitly "validate" your algorithm for applications to the proteins whose ligands are yet to be discovered.

Hi Rajeev

A RMSD calculation is frequently used to measure difference between a docked ligand (pose) with respect to its original crystallographically determined position in target protein. For example RMSD equal to zero reflects perfect reproduction of the correct pose at the original position.

Usually, 2Å RMSD for a pose is considered a docking success. Therefore, without a cocrystallized ligand RMSD calculations make no sense for docking validation. Here, one can use most straightforward method i.e. correlation between biological activity and scoring function.

Docking and crystallography are complimentary. The highest cavity in protein remains a major binding site for most of target receptors. Unfortunately, most of experts worldwide dealing with docking and scoring calculations using co-crystallized/ or highest cavity as a binding site. The structural X-ray and NMR screening studies show that a ligand binds to variety of binding sites with different binding affinity. In my opinion, other hydrophobic cavities may play important role, but are no longer chosen by experts worldwide. For its validation, correlation between biological activity and scoring/binding affinity at particular site could be used for improvement of SBDD in near future.

Hi Prashant,

How can one correlate binding affinity with biological activity and in fact there is no one to one correlation. The biological activity is dependent on the residence time of the inhibitor which in turn depends upon the stability of protein-ligand complex. I think MD simulations may provide better insight rather docking.

Good day, to our great researchers. I have tried to perform a re-dock to validate a docking procedure, but have failed in as many attempts I have made. could someone please help me with the detailed procedures of performing the re-dock and especially how to align the protein and the cocrystallized ligand using Pymol software together with calculating the RMSD value? A tutorial video will be an added effort. Your contribution will be highly needed and valid. thank you as I anticipate your urgent response.