It seems to me that for so large model is difficult to hope on success.

You can try submitting you protein sequence in Robetta 

http://robetta.bakerlab.org/

Homology modeling was it done using modeller?

@Alexey

dear Alexey

I stopeed trying to get the full model

that is why i tried to construct them as domain wise which would be around a size of 300-400 residues each. these models also still showing the problem with the orientation with membrane span. they are not lying with the membrane properly.

I want to solve this problem atleast

Hi,

            You are dealing with huge membrane protein. Even you are trying with individual domain , it will be hard to predict the 3D- structure. First try to get the information about  transmembrane region, which is necessary to know. You can do that by obtaining the hydropathy plot. Once the transmembrane region is correctly placed in the membrane, then only simulation will be stable. Robetta is an useful server for construction of ab initio models of protein, but it is not good for large proteins(>100 amino acids)

@Vikas

followed the same as you suggested

but during the simulation the structure is getting distorted out of the membrane. how can i avoid this?

Ab initio modelling with Rosetta works well for globular proteins up to 120 or so residues.  Membrane proteins can be slightly larger since, with good TM region predictions, the packing problem is simplified.  However, your 300-400 residue domains are still clearly too big.

New contact-driven modelling eg EVfold, PconsFold is worth a shot if your protein domains are parts of large families i.e. if large and diverse alignments of homologs can be obtained for them.  The key paper is here

http://www.sciencedirect.com/science/article/pii/S0092867412005090

and there is even a server you can try

http://evfold.org/evfold-web/evfold.do

Good luck.

Hi ra Nandu, try ab initio or robetta it works well.

As many others have pointed out, your model contains rather large proteins. Perhaps you would like to consider performing coarse-grained MD simulation as it may run faster. There is a software called Martini that runs on GROMACS.

http://md.chem.rug.nl/cgmartini/

They do have excellent tutorials that you may follow:

http://md.chem.rug.nl/cgmartini/index.php/tutorial

Also a tutorial on setting up bilayers:

http://md.chem.rug.nl/cgmartini/index.php/bilayers

@Daniel

your contribution helped to find some solutions. Article you shared is really excellent.

@Chanin

your answer on lipid layer addition is really good.

Thanks a lot