Dear Eric,
Thanks for the explanation
Best
Sasha
From: Eric Pettersen [mailto:pett@cgl.ucsf.edu]
Sent: Thursday, June 12, 2025 20:59
To: Alexandra Zahradnikova
Cc: chimerax-users@cgl.ucsf.edu
Subject: Re: [chimerax-users] Labeling chains in MD simulation trajectories of a multimer
Hi Sasha,
In order to read a trajectory, ChimeraX needs to know the atoms and connectivity of the system (the topology) and the trajectory coordinates. This is typically done by opening any structure file that contains the complete system (e.g. PDB, GRO) and then loading the coordinates to transform the single structure into a trajectory, though for some topology formats (e.g. PSF) it’s done as a combined step since the topology lacks even initial coordinates.
The problem with GRO files is that they only contain atom and initial coordinate information. They lack bond and chain information, so ChimeraX has to try to figure that out on its own. If atoms that should be bonded are further apart in space than the length of a reasonable bond, ChimeraX won’t bond them together, and you get the fragments you see. However, you don’t have to use a GRO file from a part of a trajectory as the basis of the trajectory — you can use any structure file that contains the topology of your system. So if you have a PDB file with the right atoms and bonds you can use that. If you have a minimized GRO that produces the right structure you can use that. You can even use a structure that is missing a few bonds and add the missing bonds with the “bond” command (probably with “reasonable false” specified) and go from there.
You might wonder why we don’t support TPR topology files, which do have bond and chain information. TPR files use an undocumented binary format. Chimera reads those files by basically reversing the complex process that Gromacs uses to write those files, and which changes with every release of Gromacs. So every new release of Gromacs requires spelunking into the Gromacs source code to determine how TPR writing changed and how to update Chimera’s TPR reading. This is a difficult, pain-staking, and time-consuming process, and we basically can’t invest that kind of time into TPR support.
I’d say your best bet is to get a PDB file of your system (possibly from Chimera) and use that as the basis for your trajectory in ChimeraX.
—Eric
On Jun 12, 2025, at 8:32 AM, Alexandra Zahradnikova via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Eric,
I would like to know more about how do you assign chain letters to the chains.
I have noticed that when the initial structure is not perfect enough, the number of chains is increasing during the progress of simulation in individual GRO files. The extra chains have usually only one or a low number of residues. I was thinking that this occurs because the residues get separated too much during the simulation (for instance it occurred when the GROMACS minimization was performed in single precision or when there were many LINCS errors during minimization). But now I have a case when the initial GRO file prepared by CHARMM-GUI has a broken chain but after minimization the resulting GRO file has intact chains which is puzzling to me.
When I open a trajectory in the old Chimera, I always get the correct number and contents of chains. I have attached a zip file with two examples, in which the files TPR and XTC opened in Chimera give a structure with complete chains A-D but the equivalent GRO file open in ChimeraX has broken files. In the PDBs I have deleted all residues except the original protein structure with ligands.
step7_1_Chimera.pdb - beginning of the simulation, TPR and XTC open with Chimera for frames 1-10, frame 1 shown
step7_1_ChimeraX.pdb - beginning of the production run, first GRO open with ChimeraX, which corresponds to the first frame
step7_1000_Chimera.pdb - end of the simulation, TPR and XTC open with Chimera for frames 991-1000 (the last TPR file), frame 10 shown
step_100_GRO_ChimeraX.pdb - end of the simulation, last GRO open with ChimeraX, which corresponds to the last frame.
step7_1000_trjctr_ChimeraX.pdb - end of the simulation, frame 1000 of the combined trajectory open after centering the protein in GROMACS.
It can be seen that (1) the RMSD between the C-alpha atoms of PDBs corresponding to the same frame is ~ 0; and (2) the last frame is still quite OK in the step7_1000_trjctr_ChimeraX.pdb, but since no GRO file in the whole series was unbroken, it cannot be repaired in ChimeraX.
Thanks in advance for feedback.
Best regards
Sasha
-----Original Message-----
From: Eric Pettersen [mailto:pett@cgl.ucsf.edu]
Sent: Thursday, April 24, 2025 02:28
To: Alexandra Zahradnikova
Cc: ChimeraX Users Help
Subject: Re: [chimerax-users] Labeling chains in MD simulation trajectories of a multimer
Hi Alexandra,
Thanks for the example file. I have committed changes to .gro file reading to do a better job of identifying chains and assigning chain IDs. In your example file, the four protein chains are assigned chain IDs A-D and all the singleton residues get chain ID E. The fix will be in the next daily build.
--Eric
On Apr 22, 2025, at 1:01 PM, Alexandra Zahradnikova <Alexandra.Zahradnikova@savba.sk> wrote:
Hi Eric
Sorry, I have sent the unfinished mail by mistake, but you got the problem perfectly.
Since I need to analyze e.g. RMSD/RMSF or H-bonds per chain, I need to have different names for the four chains. I can do it in old Chimera, because it requires a combination of a TPR and an XTC file, which will produce the correct chain names. However ChimeraX reads only the GRO file but not the TPR file, and the GRO file does not seem to have this information.
I attach a sample .gro file with the following .xtc trajectory. In the .gro file, everything including protein, ions, lipids, and water is one "principal chain" without a name. I have also added the .tpr file so you can compare the result in ChimeraX and Chimera.
Best
Sasha
-----Original Message-----
From: Eric Pettersen [mailto:pett@cgl.ucsf.edu]
Sent: Tuesday, April 22, 2025 7:12 PM
To: Alexandra Zahradnikova
Cc: ChimeraX Users Help
Subject: Re: [chimerax-users] Labeling chains in MD simulation trajectories of a multimer
Hi Alexandra,
I don't actually have access to any .gro files that have multiple chains. If you could send me an example I might be able to improve ChimeraX's behavior in that situation.
--Eric
Eric Pettersen
UCSF Computer Graphics Lab
On Apr 22, 2025, at 7:23 AM, Alexandra Zahradnikova via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi All,
I am interested in MD simulations of homomultimeric ion channels, which I do in GROMACS. I would like to analyze my output files with ChimeraX but I run into the following problem:
The .gro output of the simulation consists of a single chain named /? in ChimeraX. When I open the file in old Chimera, it shows every chain of the original protein separately but with the same name ()
_______________________________________________
ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu
To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu
Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
<step7_1.zip>
<7tdg_3668_5037_R4736W_charmm-gui-4695003116_17.zip>_______________________________________________
ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu
To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu
Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/