This query is quite popular nowadays...

Please follow the attached links

You can draw a correlation plot in between the inhibitory activity on one axis and the docking score on the another axis, this will give you the reliability of your docking algorithm. But don't expect for a strong correlation. 

Good Luck

https://www.researchgate.net/post/Can_we_predict_inhibitory_activity_based_on_molecular_docking

https://www.researchgate.net/post/Is_there_any_relationship_between_docking_score_and_biological_activity/1

https://www.researchgate.net/post/What_would_be_the_reasons_of_the_unmatching_between_docking_scores_and_IC50_values_with_references_please

 Thanks Anu,

Let me check the links.

The debatable and interesting one, however, researchers have different opinions, please go through the link:https://www.researchgate.net/post/Is_there_any_relationship_between_docking_score_and_biological_activity

Yeah Mohsin,

Thanks for your sharing of link.

It remained a crucial confusion for me for a long time. Actually, so many papers have published this kind of relationship especially in computer aided drug design avenue. What the researchers actually done, they have started their work from the IC50 value, and made some assumptions right here based on IC50. Later on they switched to docking score as detrimental parameter for activity of newly identified compounds while selecting IC50 based highly potent compound as cutting-edge reference. Now I argue that if the case is like the one discussed above, there should be a stern relationship between the IC50 and docking score, but half of the researchers are tailing this concept while half of them don't. Personally, I don't know this mechanism of tieying the two properties. If they are not very linked and the contrast is persisting, so to trust the docking result would be merely a roaming.

The main purpose of molecular docking is to get the theoretical model for the protein-ligand complexes that are not available in the database (experimentally).

People correlate the inhibitory activity and docking score in order to check the reliability of the docking protocol. 

Dear Amir Zeb,

Can we take account of the enzyme concentration, the inhibitor and the substrate along with other experimental conditions while performing docking? If not, then is it possible to correlate them?

It's true that we can not take these all factors into account while performing docking. Correlation is just meant to validate the docking protocol, it adds robustness. 

Hello dear Mohsin,

I'm working solely on computational biology. I have no access to wet lab experimentation. It's my unavoidable limitation to stick to docking only, and I should prove and  validate my claim and hypothesis while approaching the available computational resources. And you know that if the computational resources are not well confident and explanatory by itself, how may I trust their results?

As Anu is arguing that docking is performed to validate docking protocol. So for me, this is a bit shocking that docking by itself is in trial phase as her claim seems like. Let say that it is. Then how so many people have claimed and also proved their hypothesis while exploiting the docking as the decision making platform?

At the end, I can just assume that if the docking by itself is not an authentic and reliable approach, then prior to go for an ambiguous result, we should focus on the limitation(s) of docking.

Hi Amir Zeb, you are missing a point here, what I am trying to say is that when you compare docking score with inhibitory activity, you can not get 100% correlation as the basic conditions are not same, inhibitory activity, you get from experimental procedures and docking score, you obtain it through modeling using algorithms.

And when you compare the both (inhibitory activity and docking score), you will get correlation value which will interpret regarding the reliability of the docking protocol you are applying. 

Moreover, the main purpose of docking is to get protein-ligand complexes that are not present in the database or not reported yet (experimentally). So, after getting the results, you go for validation. It is not necessary to validate after getting results, you can validate before docking also. Docking gives you a value (+/-), which predicts regarding the stability of your complex. Docking score, is not responsible for predicting the activity of the ligand.

Yeah, dear Amir, Indeed we should focus on the limitations  first.

Hello Anu,

I got what you'r gonna saying, but if you kindly look at the file i have attached right to the question, it looks quiet strange.

Well, if we don't have experimentally determined structure for the complex, the only means to have a look on their mutual binding is the docking, right, now let say, we gonna taking an example and the exemplary complex fails to validate the experimental data, i mean that the docking score is pretty low for those ligands for which the IC50 value is low (highly potent) than those for which the docking score is brilliantly high and the IC50 is also pretty high (means that low inhibitory activity). Whom we should blame? Means that we can't trust docking score because the experimentally determined results could not be false or fake.

What my point of view is, I'm insisting that if the situation is persisting like the instance I already explained, the docking practice should be reviewed.

It's quite possible that I might be very wrong but I believe in actual results that we have in hand.

Also you talked about the validation of the results before and after docking. So, I got till the point that if we determined the comparative stability of a ligand in its receptor site (whatever the case is, whether this is active site or allosteric site) by docking, and we went for experimental validation and the results we got from experimentation, were in parallel with that achieved from docking, we will accept. Right. Now, let say, we screened a set of inhibitors against a receptor by experimental approach and calculated their respective IC50 values, which is directly correlated to their activity, right. We claim in docking theory that a ligand will have high score if the fitting as well as the scoring function (means derived from different components of energetic terms) are stable. And ultimately, these two parameters are directly suggests that the fitting and scoring function are binding dependent, where in IC50 calculation, the sole mean of activity is also binding, right, if some one ask, that why IC50 is binding dependent, then if there is no binding, there should be no activity. The whole discussion means that either IC50 or docking score, both of them are binding dependent, either one way or other. Well, turning back to my point, if the results we achieved from docking, were opposite to that achieved by experimental approach, then which approach we should blame? Sure, docking, because experimental data could not be faked. And this is what, I have stuck that experimentally determined results may not necessarily be validated by docking.

Anyhow, I appreciate and respect you'r opinion.

Thanks a lot.

Amir, sorry to say, but it will not work.

1) all your compounds have the same activity. IC50 is in mol/l, it is an energy property via dG=-RTlnK. To be able to see any discrimination, you need about 2 factors of magnitude. Your numbers differ b less than a factor of 2.

2) you should not expect ANY correlation, since NO docking score will give you something like an IC50.

Docking can maximally split into active and inactive with a certain quality. Dont trust publications stating they can.

And then, it will only work if your poses are correct. It sounds that you do not have a cocrystalized ligand.

I have one query.....

How did you select the pose after docking? What were the criteria's for the selection of pose?

Hello Anu,

I have focused on docking score as the selection criteria for the pose. I have compared the poses with highest docking score for both the sets whether highly active or less active

I hope you can bring me on the right path. Please take me long in this discussion, i see somehow good ending.

Thanks Anu

Dear Amir,

Anu has raised a good question about pose selection and Sir Andreas Göller  has pointed out that "it sounds that you do not have a cocrystalized ligand". Both of these quires are interrelated. What I mean to say is that one cannot compare highest docking energies of the ligands directly with experimental activity untill the interactions are not similar.  

Thanks Andreas Göller, for your answer and explanation.

I could not understand dG=-RTlnK in terms of docking. Actually i m using GOLD software for docking. Please let me know if there is any concerned literature to correlate your concern and gold docking.

I have selected a set of inhibitor of a protein from binding database  (https://www.bindingdb.org/bind/index.jsp) on the basis of IC50 values and in some other cases Ki value. They are divided into highly active, moderate active and less active nased on their respective IC50. So the inhibitors i am talking about don't have cocrystalized structure.

Thanks in deep

The only purpose of the Gold fitness score is to let one rank her/his ligands from the best fit to the worst, so that, when performing a virtual screening run, she/he could select a reasonable number of top-ranked ligands (out of a very large number of docked compounds) for experimental testing. These score are not meant for a quantitative Pearson-like correlation with any experimental metrics.