So the H-bond is not stable. Either that bound pose is not sufficiently energetically favorable or the force field parameters for the ligand are inadequate. How did you generate the complex coordinates? What force field are you using and how did you parametrize the ligand?

Dear Justin

I used GROMOS96 53a6 then I parametrized ligand by  PRODRG . In addition I have corrected quantum charge by Gaussian. I have download ligand coordinate file from "gromacs topology" link on PRODRG and added it to pdb2gmx generated protein coordinate file. Please help me find the problem.

Thanks for your support 

In general, there is only a very loose correspondence between QM charges and those used by GROMOS force fields, which are highly empirical. Without the coordinates and topology of the ligand, there's not much more to suggest in that regard. How about my first question - how did you generate the complex?  If it's just from docking or something, in the absence of solvent, polar and electrostatic interactions are overstated, so the MD behavior may actually be more realistic.

I have passed all needed steps for the complex coordinate file generation : box, solvent, ions, EM ,NVT & NPT. There is no problem with them also AutoDock results showed 6 h-bond at the binding site .

I have found 10 missing residue in my protein pdb file while I never have been notified of them by gromacs or pdb viewers. As they are not in binding pocket can they cause that problem?

Here are my ligand topology and coordinate files.

  Thank you so  much!

"No problem" with EM, NVT, and NPT is no indicator that the topology is of any quality. The application of restraints and generally the short time frame of each of these processes doesn't guarantee that anything about the ligand is right. Again, as I said before, what docking thinks is right is not necessarily correct when the system is fully solvated and thermalized.

The missing residues are only significant if they are non-terminal residues, e.g. somewhere in the core of the protein. pdb2gmx would have thrown obvious warnings about this, but do check carefully. Ignoring missing residues can completely invalidate what you're doing if they're functionally significant.

Your ligand is straightforward to parametrize using existing parameters. There's no need to do any sort of QM calculations here. The edited file is attached; please give it a try (caveat emptor: I am not responsible if this doesn't work and I have made no effort to validate that the applications of these parameters is appropriate for any purpose).  You will note that many of the charges assigned from Gaussian differ greatly from those generally used in the force field.  The charges groups also required considerable modification; yours were much too large (though that is irrelevant if you are using the Verlet scheme in GROMACS, but it will affect results with the group scheme).  For the logic behind what I did, please refer to:

http://pubs.acs.org/doi/abs/10.1021/ci100335w

ِDear Justin, I truly appreciate your help in resolving the problem.

Thank you and God bless you

Dear Justin, I have one question regarding the N terminal missing atoms in running simulations. Is there any reason in Gromacs

@K, if you have a new question I suggest starting a new topic to avoid driving this one away from its intended purpose. Or (even better), report GROMACS questions and problems to our mailing list, [email protected]. You will need to provide more details, no matter what, because your question is not clear.