[gmx-users] cannot inactivate fitting in g_rmsf - incomplete coordinates are written

[Thomas Evangelidis]

Dear GROMACS community,

I am trying to calculate B-factors of a sample equilibrated system (31
frames). My trajectory is in .psf and .dcd format but I have created a .tpr
file for my analysis from the first frame . First I eliminated random
translation and rotation of my protein by fitting every frame to the first
one (I selected to fit CA atoms and write the whole system back):

trjconv -f truncated_WT.dcd -o truncated_WT_fitted.trr -s
truncated_initial_WT.tpr -fit rot+trans

Then I tried to calculate the average coordinates and save then in a .pdb
file, either by first fitting all the frames to the first one (I selected to
fit CA atoms):

g_rmsf -f truncated_WT_fitted.trr -s truncated_initial_WT.tpr -ox -fit

or without doing a least squares superposition (I selected to fit CA atoms -
bizarre) :

g_rmsf -f truncated_WT_fitted.trr -s truncated_initial_WT.tpr -ox -nofit

However, in the latter case I am always prompted to select group(s) for root
mean square calculation, and besides that the average coordinates written to
the xaver.pdb file are incomplete. These are the first 10 lines:

==> xaver_fit_grmsf.pdb <==
TITLE     vmdmolecule0
REMARK    THIS IS A SIMULATION BOX
CRYST1  125.654  115.248  133.144  90.00  90.00  90.00 P 1           1
MODEL        1
ATOM      5  CA  MET     1      98.823  64.559 100.152
1.00408.90
ATOM     24  CA  PRO     2      98.801  61.706 101.694
1.00232.79
ATOM     38  CA  PRO     3      96.250  62.337 103.868
1.00106.03
ATOM     50  CA  ARG     4      92.760  61.521 102.890  1.00
69.93
ATOM     76  CA  PRO     5      89.888  63.926 103.105  1.00
71.21
ATOM     88  CA  SER     6      88.220  63.301 106.384  1.00
41.38

==> xaver_nofit_grmsf.pdb <==
TITLE     vmdmolecule0
REMARK    THIS IS A SIMULATION BOX
CRYST1  125.654  115.248  133.144  90.00  90.00  90.00 P 1           1
MODEL        1
ATOM      5  CA  MET     1      98.819    -nan 100.135
1.00407.71
ATOM     24  CA  PRO     2      98.801    -nan 101.677
1.00231.97
ATOM     38  CA  PRO     3      96.249    -nan 103.853
1.00105.71
ATOM     50  CA  ARG     4      92.760    -nan 102.876  1.00
69.63
ATOM     76  CA  PRO     5      89.886    -nan 103.092  1.00
70.84
ATOM     88  CA  SER     6      88.220    -nan 106.373  1.00 41.21


The same happens no matter what group I select for fitting and or which
version of GROMACS Tools I use (4.5.4, 4.0.5). I also tried to use .xtc and
.pdb file formats instead of .trr and .trp without any luck.

Surprisingly I could get complete coordinates from g_covar v4.5.4 without
fitting, without being prompted to choose a group for the least squares fit.
However, g_covar is much slower than g_rmsf and I would avoid using it for
the whole trajectory if I could. For the sake of comparison these are the
average coordinates I got from g_covar with and without fitting (file names
are self-explanatory):


==> average_fit_gcovar.pdb <==
TITLE     Average structure
MODEL        1
ATOM      5  CA  MET     1      31.396   3.841   9.383  1.00
0.00
ATOM     24  CA  PRO     2      31.375   0.988  10.924  1.00
0.00
ATOM     38  CA  PRO     3      28.823   1.620  13.099  1.00
0.00
ATOM     50  CA  ARG     4      25.334   0.802  12.120  1.00
0.00
ATOM     76  CA  PRO     5      22.461   3.208  12.336  1.00
0.00
ATOM     88  CA  SER     6      20.794   2.581  15.615  1.00
0.00
ATOM     99  CA  SER     7      19.269   4.709  18.340  1.00
0.00
ATOM    110  CA  GLY     8      21.111   5.337  21.583  1.00
0.00

==> average_nofit_gcovar.pdb <==
TITLE     Average structure
MODEL        1
ATOM      5  CA  MET     1      98.815  64.568 100.152  1.00
0.00
ATOM     24  CA  PRO     2      98.794  61.715 101.694  1.00
0.00
ATOM     38  CA  PRO     3      96.243  62.347 103.869  1.00
0.00
ATOM     50  CA  ARG     4      92.753  61.529 102.890  1.00
0.00
ATOM     76  CA  PRO     5      89.880  63.935 103.105  1.00
0.00
ATOM     88  CA  SER     6      88.213  63.308 106.385  1.00
0.00
ATOM     99  CA  SER     7      86.688  65.436 109.109  1.00
0.00
ATOM    110  CA  GLY     8      88.530  66.064 112.353  1.00  0.00


I also visualized truncated_WT_fitted.trr and average_nofit_gcovar.pdb in
VMD and concluded that the average coordinates of average_nofit_gcovar.pdb
are most likely correct (you can also work it out by comparing the
coordinates of average_nofit_gcovar.pdb, xaver_fit_grmsf.pdb and
xaver_nofit_grmsf.pdb).

Any ideas what I am doing wrong with g_rmsf?


thanks in advance,
Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110827/90a65fc1/attachment.html>