Step-by-step technical details aren't usually in most papers.  What is it that you need to know?  Both GROMACS and AutoDock have numerous tutorials available on the web.  I suggest you start with those and design your workflow based on what you learn there.

I am doing that for now.

I  need to complete my model, missing loops and refine the structure first.

Thanks

If you google for "Gromacs tutorial" you will find lots of information. Let me just highlight some points that these tutorials sometimes overlook for the sake of simplicity.

1. Decide wether you need explicit or implicit solvent. Implicit is faster, but less accurate.

2. Choose a force field. Most will be OK, but it's your choice. Several publications have reviews their relative merits. As a rule of thumb, the newer the better.

3. Most tutorials deal with holo proteins. Just normal aminoacids. If you have co-factors, post-translational modifications, metal ions, etc you have to be more careful because some automatic setup procedures can ignore them.

4. Check the protonation states of key residues. Use a server such as propKa or H++, but check the results. Mainly for residues close to your substrate. If in doubt, try two possible protontation states and see if what results makes sense (check structural stability during the simulation).

Hope this helps.

I was about to add a link to a set of tutorials that I've found particularly helpful, when I noticed they were Justin's, so I'll let him have the honour! (thanks, by the way - they're great...).

As Ramon says, the moment that you get into things that aren't standard amino acids things get more challenging... (a quick note: if you are going to be using the same non-standard acids for multiple calculations, it's probably easier to modify (once you've taken backups) aminoacids.rtp and .hdb for your appropriate force field, as well as residuetypes.dat, than to create files in each job directory and move them around.) If the protein chain only needs the natural acids, and all of your extra species are ligands, then it's an easier situation and you can use tools like PRODRG to create your topology, but check the output carefully!

For missing residues, there are many free tools available online. Depending on how many missing residues you have, it could as easy as adding a alpha carbon manually or as complicated as building a homology model/threading.

The protonation state of certain amino acid residues may also cause deviations in your docking/MD, and I would use online tools like MolProbity to deal with this issue and look into those problematic His visually.

thank you all.

I will going through these tutorials and apply what you suggest

do loop modelling if structure is having long tail on it before doing simulation by Gromacs

http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/,use this tutorial for simulation by using Gromacs

thank you all

for answers.

In my opinion you should use Modeller to solve your problem. It's enough to add missing loops in model and it is very easy. 

http://spdbv.vital-it.ch/loop_tut.html,for loop modelling by using swiss pdb viewer if long tail is there in your protein structure