i have also encountered similar situation. on further literature survey I found that

1. IC50 or Ki values are more correlating then % inhibition.

2 docking score cutoff - only enhances probability of getting the hits

3. for a given analogue series correlation may not be established may be due to solubility issues. change in binding conformation of proteins, Temparature and ionic strength of the buffer which will influence entropy. (which is often neglected in docking software)

4. very often we neglect invitro assay principal. we should consider any coupled assay involved if so what is the influence of inhibitor on second reaction should be addressed. influence of inhibitor on quenching in Fluorescent based assay, is inhibitor forming chelates with any ions which are critical for assay etc. therefore IC50 value is more reliable then % inhibition

I have listed this as possible reasons. I appreciate if any one encountered with such case shares there view in this regard

There are many differences from in-silico experiments and real-life experiments. Some you have already mentioned. Potent drugs act the same in all three cases because all the noise generated by interaction with other proteins, compounds, etc... is minimized by the affinity of the drug and the target. However, less strongly interactive drugs will tend to have more variance in the outcome. In a in-silico docking experiments all noise that it is not cause by the solution, the drug, and the protein is stripped out, and therefore you can see activity that you might (or might not) see in real life. You will need to check in your in-vitro and in-vivo assays if the drug:

1) Has been modified in any way (different PH than in the simulation?)

2) Is interacting with other parts of the protein or other proteins (the latter is less likely in in vitro if you have purified correctly the protein)

Also check if your protein:

1) had a conformational change in real solution that buried the interaction sites (PH,temp,ions)

2) have a secondary docking site that is less active (although this will be shown in the simulation)

Those are the only ideas I can come up with. Regardless, computational methods always oversimplify the problems and that is why they are only proxies. I will recommend you to try also different software before doing the experimental study, since they can give you different answers. Also re-run your in-silico experiments, making sure that the conditions (temperature, pressure, ions, and proteins present) are the same that the ones you will use in your in-vitro and in-vivo assays.

Hope this is of help.

Thanks Jose

i have also encountered similar situation. on further literature survey I found that

1. IC50 or Ki values are more correlating then % inhibition.

2 docking score cutoff - only enhances probability of getting the hits

3. for a given analogue series correlation may not be established may be due to solubility issues. change in binding conformation of proteins, Temparature and ionic strength of the buffer which will influence entropy. (which is often neglected in docking software)

4. very often we neglect invitro assay principal. we should consider any coupled assay involved if so what is the influence of inhibitor on second reaction should be addressed. influence of inhibitor on quenching in Fluorescent based assay, is inhibitor forming chelates with any ions which are critical for assay etc. therefore IC50 value is more reliable then % inhibition

I have listed this as possible reasons. I appreciate if any one encountered with such case shares there view in this regard

Vadiraj those are interesting points,

Thank you so much Mr. Hleap and Mr. Kurdekar for your enlightening thoughts and fruitful discussion. It would be perfect if you could kindly provide some references supporting the provided ideas.