The enhance the correlation between the in vitro and in silico results you could conduct consensus docking (compare the RMSDs of the best solutions with two or more softwares), molecular dynamics and MM/GB(PB)SA calculations.

You could also change the docking protocol (scoring function, grid space, active waters, flexible side chains) to improve the docking results:

Article OPTIMIZING MOLECULAR DOCKING PROTOCOLS OF PYRROLE CONTAINING...

Article Molecular Docking Study on Galantamine Derivatives as Cholin...

Computational modeling of molecules is based on the theory after simplifications and assumptions. Sometimes the error hides in the approximation level of the computation and sometimes it hides in the user pre- and post-processing. Do not forget, that molecular docking is typically performed on a rigid macromolecule.

Some of the pesky problems that are not addressed by the in-silico analysis are related to the behaviour of the compounds in aqueous solutions: poor solubility of the compounds, aggregation, unspecific interactions with cell surfaces and sample tubes.

Multiple reasons. Docking with one protein and assume active ligand, But practically more other protein also involve it maybe one reason and also need M. D studies. Still also need pharmacokinetic parametrs to be analyze for final prepration otherwise in-vitro Or In-vivo active but failsd due to solubility, dissolution, bioavailability.

Dear Ahmad Dzulfikri Nurhan,

If you have reliable experimental results, you have a very accurate guide for your docking calculations that will give correct results and be consistent with the experiments even if you perform blind docking. Since you have experimental results, you can perform site-specific docking. It is not true that the protein is always frozen. Nowadays you can perform docking simulations with selected flexible residues.

Your question is not specific. You should describe the problem in detail. From your data, I cannot even imagine if there is a mistake in your experimental or computational approach and practice. To this end, I disagree with some answers above. There are very reliable assessments of the computational tools (that must be studied) and then you must choose the method, the tool, and the approach to address a scientific problem. I think that nobody can answer your question with the data given and it should not be answered at all.

I would love to help you though:

If I were you, I would do the following:

First, check your experimental procedure. Check everything and repeat the experiments step by step. It is your best guide and the ultimate judge. If something is not confirmed experimentally, it is just a simulation. (Few people have the knowledge to guide the biochemical experiment through simulations. And they must know biochemistry and chemistry.) I believe you are at the first stage of your scientific career, or you enter the field of docking now. So, make sure that your guide (experiment) is reliable.

Then read theoretical papers regarding the assessment of docking tools and approaches. What tool is appropriate for your research? What is the best? Do you have it? Also, detailed tutorials will help.

Third, perform a very exhaustive and systematic search for docking by altering the parameters.

If you need an answer to the question, describe the problem in a way that we can reproduce the whole procedure (experimental and theoretical).

If the steps above do not work, then you must dive into the structural biochemistry of the protein and address parts of it that may be IDRs (something that demands different approaches). I just named one potential parameter.

In pharma, 90% of the new potential drugs entering the final stage of clinical trials are withdrawn. Of the rest 10 that reach the market, 50% are withdrawn because of side effects. One reason for this problem is docking.

Lately, a lot of people work with computational chemistry tools but they lack knowledge either in chemistry or solid theoretical methodology.

More and more theoretical papers are found to be irreproducible. While in docking you cannot expect full reproducibility because of the nature of the tool, irreproducible quantum mechanical calculations are unacceptable.

Since you are a new kid on the block try to gain deep knowledge of what you do.