Thats a intriguing question. To put in simple way

First, Make sure whether ur docking methodology can reproduce the ligand pose of the crystal structure. This gives confidence that the docking methodology works.

Rather than choosing lowest binding energy, choose ligands which have more conformations in the largest cluster (i prefer to run 100 LGA runs in AutoDock and choose the ligands which binds 100 time in largest cluster out of 100 runs)

Also, Make sure that the ligands make hydrogen bonds with important residues, this is an directed and knowledge-driven approach.

Take care of ADMET properties of the ligands also.

Hope this helps.

Hello,

I usually follow these steps,

1- If you have some of the already known ligands than go for ADMET properties prediction for bot known ligands and your test ligands, see if the range mathces or you can go for PCA analysis.

2- Follow the steps of Parmeswaran Saravanan Sir

3- Observe the nature of active site (Hydrophobic or hydrophilic) and observe how many H-bonding or Hydrophobic binding occurs

Hope I was able to make my point clear and hope it helps.

Thank you somuch for your (Parmeswaran Saravanan and Ankur Sir) kind reply still I am confused with your answers so here I explain my work briefly to facilitate our discussion.

OBJECTIVE: To find targets for novel ligands.

First, I have done ensemble docking between four of my novel individual ligands with all structures which is present in PDB database by using Ensemble docking softwares.

After that I have selected 13 targets based on Pathway, disease , biological specific and lowest binding energy.

Then, I continued with single docking(Autodock vina) to confirm ensemble docking results where I kept both ligand and receptor (binding site) as flexible ,1000 LGA runs and spacing ~0.275-0.375 A .

The above single docking step I have done for both of co-crystal and novel ligands which is belong to same targets In order to compare the binding energy between co-crystal and novel ligand. End of this step, I have selected ligands which gave more binding energy and Hbonds than Co-crystal ligands.

Finally, I want to suggest four of targets to do wet lab experiment with my ligands based on binding energy,Interactions (Hbond, Hydrophobic,Pi-Pi,Pi-cation) and ligand poses completly fit with co-crystal ligand poses, Is that a good decision ? or can you tell me If any other In silico techniques to give more confidence to continue wetlab experiment ? Is it neceessary to do MD simulation ?

I think, your questions/comments are little confusing. You have selected 13 targets (Targets or ligands?)

I am assuming, you have 13 ligands. In such case, I think, it should be easy to work with ONLY 13 ligands. However, if you want to select only 4 out of 13, then first I think, you should start with clustering. Select the diverse compounds to start with.

I my view, though

Its good to rely on the docking energy but its always good to support ur finding with different tools and softwares; you have to keep in mind that VINA has a -2.85 kcal/mol standard error, so rank your molecules accordingly. MD is the most suitable for that, but before that it would be beneficial to go for QSAR analysis.

for eg. if a ligand is having more LogP value and ur predicted target is say hydrophobic in nature then this ligand may have better chances to interact and in this case H-bonding factor alone should not be the only considerable factor but u also have to consider hydrophobic interaction.

al the best..

Thank you somuch for your kind reply (Pankaj Daga) sir

I have selected 13 Targets after Ensemble docking because here my main aim is to finding the target so I have docked four of my novel ligands with All known PDB structures present in PDB database.

secondly, I have done single docking to confirm the results of Ensemble docking and to differentiate the interaction of my novel ligands with Co-crystallize ligands.

After this single docking step, I have selected only Four targets based on binding energy, interaction, good ligand (Novel ligand and Natural ligand) pose fit.

Finally, I want to do wet lab experiment to confirm that really my ligands are interacting with those four targets or not ? and we knows, it is expensive to do.

My main question is that after docking,Is their any insilico technique to give more confidence for suggesting to do wet lab experiment ? and Is it necessary to do MD simulation ?

ENSEMBLE DOCKING

(4 ligands ----> ~5000 PDB structures (targets) )

|

SINGLE DOCKING:

(Co-crystal ligands -----> 13 targets)

(4 ligands -----> 13 targets)

|

One Targets for each ligand (total four targtes)

|

???????

(Is thier any Insilico technique to give more confidence for suggesting wetlab )

(Is it necessary to do MD simulation)

|

Wet Lab experiment

To confirm the interaction between 4 Targets with respective Novel ligands

Indeed, MD simulations will give you more confidence and even you can use MM/PBSA method combined with the simulations to re-score docked ligands.

There are many publications, where such re-scoring schemes remove false postives. For example. you may read following papers:

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0051329

http://onlinelibrary.wiley.com/doi/10.1111/j.1747-0285.2009.00780.x/abstract;jsessionid=EE27013E95B9F9E81C71B3307170C043.d02t02

Yes, MD would be preferable way to re-score docked complexes and to remove false positives as Rajendra mentioned.

Thanks all for your kind replys

Dear Chintamani,

when I did ensemble docking for one ligand (against ~5000 structure) it took two days to finish and I gave input format as PDB code because I just wanna see that whether my assumed targets having high binding energy ?

Hi Mahesh,

After you performed the docking and got the protein-ligand complexes with minimum energy and high affinity you can simulate them further to check which of the complexes would be stable in the solvent system as docking is done in the vacuum conditions you will have to validate it in a solvent system mimicking the behavior of the complex in the body like conditions. Hope this helps.

Regards