I haven't used AMBER in a long time, but I think "strip" is the command you want, i.e.

strip :100-105

strip :250-255

See http://archive.ambermd.org/201110/0196.html for an explanation of caveats to removing parts of your structure; maybe that information is relevant. You should probably ask your question on the AMBER mailing list; it's a very active and helpful community.

The following should do the job for you.

trajin prod.mdcrd

trajout 100ns_center.mdcrd

center :1-400 mass origin

image origin center familiar

rms first out 100ns_centeredtercut.rms :1-99,106-249,256-400

@Justin @Senthil Kumar: Thank you so much for the answers. Both the methods worked.

The rmsd that I am getting is attached herewith. I am unable to decide if the sampling is sufficient and if the system is stable. I have performed the simulation for 150ns.Even after removing the highly flexible loop regions from the mask, still I end up with a similar rmsd pattern with fluctuations of about 1 A. Will this rmsd plot be accepted?

Kindly provide your valuable suggestions in this regard.

Thanks in advance.

The RMSD does not look particularly stable to me.

Did you use C-alpha atoms or all-atoms. the system seems to continuously deviate from the first frame. can u use reference PDB and see. I am not sure if that can help in finding the problem.

For reference pdb based rmsd calculation:-

reference xray.pdb

rms reference mass out 100ns_centeredtercut.rms :1-99,106-249,256-400

better take reference of minimized structure that you must have achieved before production run

trajin final.mdcrd

autoimage

reference minimized.rst [minrst]

rms :1-99,106-249,256-400 ref [minrst] perres perresout residue_rmsd.dat perresavg residue_rmsd_avg.dat range :1-99,106-249,256-400 perrescenter

and if you want to get RMSD of backbone change last line to :

rms :1-99,106-249,256-400@C,CA,N ref [minrst] perres perresout residue_rmsd.dat perresavg residue_rmsd_avg.dat range :1-99,106-249,256-400@C,CA,N perrescenter

@Justin : Wat would be your suggestion in this regard ? Should I extend my simulations? Or repeat the simulation?

@Senthil: I have used CA atoms only.

@Senthil and Nitin: I have tried using energy minimized structure as reference as well. It did not make any notable difference. This is a docked system and so I do not have crystal reference.

Kindly provide your valuable suggestions.

Thank you in advance.

Watch the trajectory and see what's happening. You have sudden jumps after about 120 ns, which can indicate structural transitions or instability. Extending the simulation seems most logical.

Did you repeat the simulation? As @justin said you can strip trajectory after 120ns but that will make more sense if same pattern is found in 2 or more trajectories

@Justin I shall extend the simulations in that case. But I have looked at the trajectory and the change seems to be so gradual ( when I monitor it frame by frame) that I am unable to precisely locate it.Does this plot appear skeptical even if there is an inherent bending and/or twisting movement that my protein displays throughout the trajectory?

@Nitin: I had previously done this simulation in gromacs with a different forcefield and I have observed a similar trend in the plot. I have not yet repeated the same system with the same forcefield.

@revathi though you have performed the same simulation and observed the similar trend is a good idea what to expect BUT you can't compare two simulations with different force fields as far as research paper goes.

On the contrary, comparing two simulations under different force fields is an excellent idea, especially here. Doing so demonstrates that the outcome is not biased by the force field. Reviewers often question the choice of force field, and if one can prove that the outcome is independent of which force field was chosen, the arguments made are stronger.

To the larger point, neither I nor anyone here can try to interpret this outcome for you. It's something only you can do because you have the trajectory and can do the analysis. Simple RMSD calculations are trivial, but often do not tell the whole story. To elucidate the reasons behind what you're doing, you will have to spend more time rationalizing what is happening quantitatively. If there is bending/twisting in the system, you should be able to calculate the extent to which that happens, and then further decompose that into which states are sampled at what times during the trajectory, thus explaining any conformational transitions that might be occurring.

thanks @justine for your views on use of two force fields as part of MD simulation.

Dear Amber users,

Interesting , I will start with some RMSD calculations using Ptraj in amber and hopefully I found this question on the Research gate.

My question is the following: I am trying to analyze some trajectories performed by somebody else. In such system I have one protein that has two chains (267 residues ach one) , even though there are two chains they are named equally as "Chain C" , so I open pymol or VMD I can see two chains numbered from 1-267 residues each one.

Now that I am about to start RMSD and RMSF calculation I wonder how could I make the calculation for each one of the chains separately in order to compared them. Can anyone tell me how can I run this calculation for RMSD and RMSF for each of the chain proteins?

Seems to me that is not enough to rename one of the chains of the pdb, right? because .psf, dcd files does not contain this information, all the simulations were performed considering the two chains as "C", or may I need a new topology file?

Thanks in advance, I would appreciate any comment or suggestion,

Marlet

so both chains have numbering from 1-267 and labeled as chain C?

Yes, both chains are labeled as Chain C

And what about their residue numbers? If the residue numbers are different then you can use mask :1-267 and :267-534

Their numbering for the two chains , named "C", they are numbered from 1-267 each one. I do not know whether I can change the numbering the .pdb, but the proble is .psf, all .dcd and simulation files have been done using that numbering. Therefore I do not know which is the best solution to do this.

@Justin : I have extended the simulations up to 250 ns till now and still I see nearly the same type pf pattern in rmsd.

I have also quantified the angle of bending in the protein. It is a 10 degree fluctuation and varies continuously across the trajectory. I have attached the image herewith.

Can this angle of bending be the reason for my unstable or rather continuously fluctuating RMSD?

Moreover, the transition between the bending states the protein is showing is very drastic, isn't t? From your experience has any other protein shown such large fluctuations in bending dynamics?

Kindly provide your valuable suggestions.

Hi

Firstly, when u say bending. is it between domains. which region was used in calculating the angle.? That will explain the validity of the analysis.

Secondly, do check if ur protein has intrinsic disordered regions. a simple RMSF and further supplemented by secstruct (secondary structure) analysis in ptraj will tell u the clearly the location contributing to the high RMSD.

@Senthil : My protein is a pseudo dimer and the bending happens between the monomers of he dimer. So the angle was calculated with three points: one in each monomer and third at the hinge region.

The protein does not have disordered regions. And the regions with high flexibility are the distal loops. Even if I perform the rmsd calculation removing the loops, the same pattern is seen.

If the RMSD is simply oscillating, that's not a problem. The original plot showed a noticeable increase at the end of the run.

Regarding the bending angle, what is the angle you are measuring? A fluctuation of ± 5 degrees is very small for just about anything; protein domain motions can cause much larger bending.

@Justin after extending the simulations: Please find the attached final RMSD plot. Could you tell me which part of the trajectory is best suited for analyses?

And the bending angle is between 3 points: Center of Mass of N-terminal domain, COM of the hinge region and COM of the C-terminal domain

Please find the attached RMSD plot

Analysis should be done on the converged part of the trajectory. Block-averaging of RMSD is one way to start determining convergence, but it's not the be-all end-all of criteria.

As for the bending, that's a relatively small domain motion, but centers of mass can often be somewhat crude metrics, anyway.

@Justin : If COM is a crude way, is there any other way ? I was of the assumption that choosing an arbitrary atom was more biased and predetermined the COM approach was less biased?

Kindly provide your valuable suggestions in this regard.

It is indeed less biased, but what I mean by "crude" is that it may not be very sensitive, so you may not see large changes. Consider the following: two domains rotate relative to one another, but the structural change does not strongly affect the COM; there's an obvious motion that is still difficult to quantify in this way.