Hi,
Currently clustering can only be performed with a single topology due
to the underlying way coordinates data sets are handled.
> I tried also with stripping the trajectories with mask "!(.C,CA,O,N)"
> however I suspect that with this approach I won't be able to retain full
> atom representative structures.
You're thinking along the correct lines here. Perform the clustering
with this mask, then you can do a second pass where you pull out the
representative structures with 'trajout onlyframes'. The only wrinkle
is that you need separate 'trajout' statements for each topology. For
example, say you do the clustering and end up with representatives
being frames 4,37,87,112. Frames 4 and 37 belong to topology1.parm7
and 87 and 112 belong to topology2.parm7. You could then do a second
pass (using of course the exact same input trajectories) with
something like:
parm topology1.parm7
parm topology2.parm7
trajin mytraj1.nc parmindex 0
trajin mytraj2.nc parmindex 1
trajout myreps.mol2 multi onlyframes 4,37 parmindex 0
trajout myreps.mol2 multi onlyframes 87,112 parmindex 1
Hope this helps,
-Dan
On Mon, Feb 13, 2017 at 7:59 AM, Jan Ludwiczak
<j.ludwiczak.cent.uw.edu.pl> wrote:
> Hi,
>
> I have the following setup: 20 models each having N short independent
> simulations. Models differ in the amino acid composition (all atom
> topologies are different) however their length is the same (backbone
> topologies match each other).
> I would like to perform combined clustering on the backbone atoms of all
> trajectories for all models and obtain all atom representative structures.
> Above approach works if I have only one model when I use the following
> CPPTRAJ input:
>
> parm A.parm7
> trajin 1.nc
> trajin 2.nc
> (... this goes N times...)
> rms first .C,CA,O,N out A.rmsd
> cluster out ... summary ... info .... repout .... repfmt pdb averagelinkage
> clusters ... rms nofit .C,CA,O,N
>
> However I wonder if it's possible to use cpptraj to perform similar
> analysis on e.g. 20 models. Here's what I worked out so far:
>
> parm A.parm7 [A]
> trajin A1.nc parm [A]
> trajin A2.nc parm [A]
> (... this goes N times...)
> parm B.parm7 [B]
> trajin B1.nc parm [B]
> trajin B2.nc parm [B]
> (... this goes N times ...)
> (... this goes for each model)
> rms first .C,CA,O,N out A.rmsd
> cluster out ... summary ... info .... repout .... repfmt pdb averagelinkage
> clusters ... rms nofit .C,CA,O,N
>
> Which results in the: Error: # atoms in current topology (1429) != # atoms
> in coords set "_DEFAULTCRD_" (1449) (printed for each model with diffreent
> atom number).
> I tried also with stripping the trajectories with mask "!(.C,CA,O,N)"
> however I suspect that with this approach I won't be able to retain full
> atom representative structures.
> Concluding, I would like to ask whether it is possible to perform this kind
> of analysis in CPPTRAJ in a single run.
>
> Many thanks in advance,
> Jan Ludwiczak
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Feb 13 2017 - 06:30:03 PST
