[gmx-users] superimposed structures

[Mark Abraham]

leila karami wrote:
> Dear Mark
>  
>  
> *(1)* I did following steps for obtaining superimposed structure:
>  
> I obtained 0.gro and 5000.gro files by above command                 
>                                                                      
> trjconv -f *.xtc -s *.tpr -o n.gro -dump n-1                         
>                                                                      
> I concatenated 2 input files ( 0.gro and 5000.gro).                  
>                                                                      
> trjcat -f 0.gro 5000.gro -o f.xtc                                   
>                                                                      
> finally,                                                             
>                                                                      
> trjconv -f f.xtc -s *.gro  -o f.pdb -ter -fit progressive -n n.ndx
>  
> simulation time was 5000 ps.

That works OK, but there's no need to use .gro as an intermediate 
format, or need trjcat. If your trajectory is only the 0-5000 segment,

trjconv -dt 5000 -f in.xtc -o first_last.xtc

just works. Then

trjconv -f first_last.xtc -s the.tpr -o first_last_fitted.pdb -ter -fit 
progressive -n n.ndx

Actually you can probably just use

trjconv -f in.xtc -s the.tpr -o first_last_fitted.pdb -ter -fit 
progressive -n n.ndx -dt 5000

to do it in one hit.

You can learn quite a lot by reading trjconv -h (as I write on this list 
about once a day...)

> *(2)* I used command : [g_confrms -f1 initial.gro  -f2 final.gro -n1 
> 1.ndx -n2 2.ndx -o fit.pdb -fit] 
>  
> to obtain superimposed structure of initial and final structures. I saw 
> fit.pdb file by vmd but fit.pdb obtained was very mixed,busy and unclear.

That's inevitable. The contents of the PDB file contain only sets of 
coordinates. VMD then guesses where bonds are based on the geometry. If 
there's two copies of each atom quite close to each other, then there 
must be a pile of apparently spurious bonds. Like I said last email, if 
you feed something unrealistic to VMD, you shouldn't be surprised when 
it makes poor guesses about what is going on.

I guess what you want to do is align your two structures (e.g. trjconv 
-fit as above), separate them with trjconv -sep and open both in VMD 
separately. Now they'll be aligned, but independent from the point of 
view of VMD. While viewing, if you want to rotate them together rather 
than separately, read the VMD manual to work that out.

Mark