Hi, Dammar, i think you have different situations in there.

1) if you do not have a previous experience with MD, I strongly recommend you start with a simpler system, perhaps the isolated units of yours system. Everything will be faster and simpler for a monomeric and uncomplexed protein, from the simulation setup to the analysis;

2) you should read "a little" of the literature prior to choosing the force field and water model, as every result you obtain will be dependent on this choice;

3) attaching a new domain to a protein is something delicate, but to 8 simultaneously is at least tricky (it is usually done with docking). How you define the relative orientation of the new domain to the previous one? This is not a matter of simply putting the domain there, but to establish a reliable relative orientation between domains. Normally in MD you do not have enough simulation time to access inter domain movement, and in a huge system as yours, you will be kept in smaller time scales, seeing more local than global movements. Then the starting conformation will determine everything in your simulation;

4) the error you are obtaining seems to be related to how chimera saved your pdb file, using non-standard names for the residues, as compared to how they are defined in the force field. You may check the residue name, in relation to what is defined in the rtp, or chose another program to do that - assuming that the strategy you employed in chimera is adequate for what you want to see.

Sometimes we (or others, as the advisers) want to model molecular systems as they occur in a given experimental or biological condition. Whereas at each year this becomes feasible for more situations, it is still at least challenging for most systems. Reduce the those systems to a minimum representative composition usually is a good starting point, isolating aspects and molecular events. It will be easier to setup things, understand errors and perform the relevant analysis.

Regards,

Hugo

Well, obviously, the initial pdb structure you are trying to use has some problem for OPLS force field (ff). Generally, if you need to use OPLS ff, and want to generate topology in Gromacs by pdb2gmx, you should have the right type of atom naming as the OPLS ff. Just like the Met1 HB3, is not a standard name for hydrogen in OPLS, but might be recognized by other ff.

So suggestions here: 1) In order to use OPLS, you firstly need to check the intial PDB file and modify the atoms' names according the OPLS ff for amino acids. then you could use pdb2gmx and choose OPLS ff.  2) before set up topology, check missing atoms and add them (using deepview for example) 3) you could find a tool to generate topology, instead of using pdb2gmx in gormacs. Life will be easy to use another software to generate topology file other than Gromacs. For example, you could use the xleap and tleap tool in AmberTools (free software). 4) you have so many subunites, actually, you also could build topolgy for the 3 types seperately, and then assemble them to the same topolgy file.

As for choosing force field, I am always fond of Amber force fields, though the topic sometimes is a problem of personally taste, but different force fields do have differences. Like OPLS force field more is biasing torwards helices structure. The reason I prefer Amber force field is that Amber force field is easy to use in AmberTools and Gromacs, and I could generate topology in AmberTools, and then convert into Gromacs topology format using a Amb2gmx.pl script.

And before you run simulation, you need to double check wether there are some missing atoms, and the protonation state of some key resides (HIS,HIE,HID). And also take care the disulfide bonds between CYS residues.

Good luck! 

Thank you Hugo and Liangzhen.

Liangzhen,

I will try the Amber force field. Previously I had tried OPLS. Sorry, I forgot to mention you that I had also used CHARMM force field.

Hugo,

I have tried for single subunit with attached GFP and it worked well, for that I used CHARMM force field. I had made the pdb from chimera compatible with CHARMM force field using CHARMM-GUI( http://charmm-gui.org/?doc=input/pdbreader). It worked for single subunit but not for whole structure.

Regarding the use of chimera for attaching the GFP on each subunits, how I did this is, first I use single subunit attached with GFP and by aligning one by one got the complete structure and saved it as pdb. So, the original orientation of 8 subunits is unchanged but I am not sure whether or not the orientaition of attached GFPs matters?

Thanks alot.

Dammar

For selecting suitable force field for protein-ligand complex, just make sure to check the literature if someone has already performed simulations on same receptor. This will save your time and if not, then go for AMBER or CHARMM ff.