AMBER Archive (2009)

Subject: Re: [AMBER] Very weird topology problem

From: Peter Gannett (pgannett_at_hsc.wvu.edu)
Date: Sat Jan 24 2009 - 17:17:26 CST

I am having a problem that sounds like it might be related.
 
1) I have a trajectory, solvated. It is a DNA with a modified base.
A pdb made with ambpdb from the restart file and topology file (fully
solvated) look just fine. xleap is perfectly content with it, no
missing bonds, etc.
 
2) I take the solvated pdb, edit out the waters/ions and make new
topology and coordinate files.
 
3) I run ptraj - cluster option to find the most representative
structure and make a pdb with the topology file made in (2). When this
structure is loaded in xleap (pymol....) the connectivities of the
modified base are all screwed up and some of the atom names seem to be
wrong though.
 
I compared the modified base in the pdb with waters/ions with the pdb
in 3. Some of the atom names have been moved around. If I rearrange
the atom names (and only the atom names, not the coordinates) of the
file in (3) so that they are as in the pdb file in (1), the problem is
fixed.
 
I have been trying to carefully work through all of this to find
exactly where the problem gets introduced but so far can make no sense
of this.
 
Pete Gannett

>>> On 1/24/2009 at 4:14 PM, in message
<BAY110-W4E1366730126F1B4D16DFC7CC0_at_phx.gbl>, ranqi zhu
<zhuranqi_at_msn.com> wrote:

Hi, Everyone,

I am having a very weird problem.
I have some trajectory from gromacs simulations and I want to use ptraj
in amber10
to cluster them. So the raw idea is:
1. extract pdb from raw xtc file (I can extract amber trj (g87 format)
directly actually, but the unit is different from amber which cause
trouble, so I extract pdb instead)
2. I need to generate a topology file that is needed by ptraj. I have
to use tleap to do this.
So I pick a random pdb from the result of step 1. Since there are some
naming convention difference between gromacs pdb and amber pdb, i have
to manually change a few of them (For example, HB1,HB2->HB2,HB3). Then
the tleap works and I can get a amber topology file.
3. Then using this topology file, I can use ptraj to convert all the
PDBs I get from step1 into amber trj files. Then concatenate all the
single trajs into a big traj and cluster them. (I did delete first line
of single trjs and add a first line after concatenation).
4. I got partially broken representitive structure from the ptraj
clustering. The strucutrue shows missing bonds in vmd or pymol. It feels
like vmd or pymol has the upper limit for bond length and some atom
distance are a little over that, so vmd/pymol refuse to display them.
However the oringall PDB from step 1 looks perfect. Representivive
strucutre generated from ptraj should be directly chosen from the
original structures from step 1 and shouldn't have this problem. So I
think the problem is from the pdb->trj procedure.

So, I did some test to pinpoint this problem.

step 1
I geneterate the topology file by the
tleap -f leap.in

the content of leap.in is:

source leaprc.ff99SB
model = loadpdb p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb
saveamberparm model conf_amber.top conf_amber.crd
savepdb model conf_amber.pdb
quit

step 2
Then I have this conf_amber.top. Then I can do the convert by
ptraj conf_amber.top <traj.in>traj.out

The content of traj.in is:

trajin p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb
trajout p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.trj trajectory
step 3
After doing this I got a trj file. Then I convert it back to pdb and
problem happens

ptraj conf_amber.top <trajpdb.in>traj.out

The content of trajpdb.in is :
trajin p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.trj 1 1 1

trajout p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb pdb

step 4
after step3 we get a pdb which is
p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb.1

use pymol to compare , pdbs and topology are attached.
p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb,
p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb.1

p53TAD~wt13~RUN0~opls-25-Trjs-protein3654.pdb.1 miss bonds just like
the representive structur
es from
generated by ptraj clustering.

Any idea? The exact same procedure from all the other simulations. Only
this one have problem.

Best,
Ranqi

_________________________________________________________________
Windows Live™: E-mail. Chat. Share. Get more ways to connect.
http://windowslive.com/howitworks?ocid=TXT_TAGLM_WL_t2_allup_howitworks_012009
_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber