Thanks for the help! I have parametrized the complex using ANTECHAMBER and it was showing the net charge of +0. I have used the charge method AM1-BCC.

What are you looking for by simulating your systems? If you expect that your system behaves as you wish/like, so why do you waste your time and also your CPU sources by simulating?

It might be possible to find out the reason of the 'instability' by visualizing the trajectory. Are there any water molecules competing with the ligand for the binding site? Is there any local conformational change in the protein that triggers the unbinding of the ligand? Are there any interactions between the ligand and the protein that can be easily tracked during the simulation (such as hydrogen bonds) and that break down?

Have you tried to calculate the protein-ligand binding free energy----if yes, do you get a good agreement with corresponding experimental data? Do different ligand poses obtained from molecular docking calculations result in similar protein-ligand binding free energies in subsequent simulations? Have you tried to vary the initial conditions for each ligand pose, for example the random seed for the generating of initial velocities, and running, for example. 4 x 20 ns, rather than 1 x 80 ns, for each pose?

It is generally better to run shorter MD simulations and explore the configurational space near the initial coordinates, than to let the system drift away uncontrollably. (i have done well over one thousand MD simulations, and none of them was longer than 10 ns.)

As the answer by Ramin implies, the break-down of the complex does not mean that the stable part of the trajectory cannot be meaningfully interpreted.

You could try several resulting poses from your docking calculations as a starting point for multiple MD simulations. The first pose does not have to be the"BEST" pose.

On the other hand you should check your MD protocol. Did you perform a long enough equilibration stage?

Also, you could start your MD with some ligand protein constraints like those provided by potential hydrogen bond interactions and continue lateral with a constrainless MD.

Thank you all for help! I will get back to you once I'm able to come up with the results.

Of utmost importance is that your starting structure (docking pose) should be proper. You may want to spend more time there to get the best pose. Also, could you elaborate on what you mean by 'detaching'? Quantifying it in terms of RMSD will be more helpful.