Project

General

Profile

Bug #382

g_mindist writing bogus information in atom-pairs file when -or is invoked

Added by Shay A over 9 years ago. Updated about 9 years ago.

Status:
Closed
Priority:
Normal
Category:
analysis tools
Target version:
Affected version - extra info:
Affected version:
Difficulty:
uncategorized
Close

Description

Hello,

I used g_mindist analysis as follows:
g_mindist -f *.xtc -s *.tpr -o atom-pairs.out -od mindist.xvg -or res_mindist.xvg
Surprisingly, in the atom-pairs.out only several atoms in one
residue (from group 1) were at minimum distance from group 2
throughout the entire simulation. Checking this output manually showed
it to be incorrect.
res_mindist.xvg seems to hold the correct data.
When running the same analysis, only omitting the -or flag, we get a
correct atom-pairs.out.
g_mindist -f *.xtc -s *.tpr -o atom-pairs.out -od mindist.xvg
Comparing mindist.xvg from both scenarios shows no difference (files
are identical).
I suspect there is some bug whenever the -or flag is invoked. Strangely enough, it affect the .out file. 
Supplementary details:
1. The same thing happened when g_mindist -s was supplied with either
.gro, .pdb or .tpr files.
2. Index-groups 1 and 2 consisted upon full molecules with no
ommisions (full protein and not only C-alpha).
3. We ruled out PBC by using -nopi
4. This is common with the following Gromacs versions: 3.3.3., 4.0.5,
4.0.7.
(42 Bytes) Example for g_mindist under gromacs 4 Shay A, 06/10/2010 10:42 AM
(44 Bytes) example for g_mindist under gromacs 3.3.3. Shay A, 06/10/2010 10:45 AM

History

#1 Updated by David van der Spoel over 9 years ago

Can you please provide some input files to reproduce this?

#2 Updated by David van der Spoel over 9 years ago

Sorry I can not reproduce this bug. With and without the -or flag I get the same results in atom-pairs.out.

#3 Updated by Shay A over 9 years ago

Created an attachment (id=463)
Example for g_mindist under gromacs 4

Dear David,

Below I give the list of files, and exact commands I used to help recreate the situation.
In both cases (gromacs 3.3.3 and 4.0.x) The atompairs files do not match (the middle column is wrong).
I have included the following files (input and output) for your inspection:

<<< UNDER GROMACS 4.0.7 >>>
  • grom4sample.gro sample gro file of the system (time = 0ps).
  • grom4sample.xtc sample xtc file, 1ns simulation
  • grom4sample.tpr
  • topol.top Topology of the system
  • index.ndx Index file of the system. I used group number 1 (Protein) and 12 (EIC) for the demonstration.
I included the output for grom4sample for g_mindist with/without the -or flag. Command lines as follows:
(WITH -or flag)
g_mindist -f grom4sample.xtc -s grom4sample.tpr -n -o atompairs_with_or.out -od gmindist_with_or.xvg -or res_with_or.xvg
OUTPUT file names
  • res_with_or.xvg
  • gmindist_with_or.xvg
  • atompairs_with_or.out
(WITHOUT -or flag)
g_mindist -f grom4sample.xtc -s grom4sample.tpr -n -o atompairs_NO_or.out -od gmindist_NO_or.xvg
OUTPUT file names
  • gmindist_NO_or.xvg
  • atompairs_NO_or.out

The atom-pairs shown in atompairs_with_or.out are incorrect. I did the same for gromacs 3.3.3. ( the link for the gromacs 3.3.3 example is: http://ibis.tau.ac.il/hemi/1nsgrom333.tar.gz ).

<<< ---------------- >>>

Regards,
-Shay

#4 Updated by Shay A over 9 years ago

Created an attachment (id=464)
example for g_mindist under gromacs 3.3.3.

This is an example of input/output I made for gromacs 3.3.3. The occurrence here is the same: The middle column of the atompairs.out file is wrong if you include the -or flag, and correct if you do not.

Regards,
-Shay

#5 Updated by Shay A over 9 years ago

(From update of attachment 463)
Wrong output for atompairs.out file if -or flag is included.

#6 Updated by Shay A over 9 years ago

(From update of attachment 464)
Wrong output for atompairs.out file if -or flag is included.

#7 Updated by David van der Spoel over 9 years ago

Your file is > 200 Mb, can you please make a shorter trajectory?

#8 Updated by David van der Spoel over 9 years ago

Nevermind, I have reproduced your report now, will look into it.

#9 Updated by Shay A about 9 years ago

(In reply to comment #8)

Nevermind, I have reproduced your report now, will look into it.

Hello David,

Is it ok for me to remove the following links (the example-links)?
http://ibis.tau.ac.il/hemi/1nsgrom4.tar.gz
http://ibis.tau.ac.il/hemi/1nsgrom333.tar.gz

Regards,
-Shay

#10 Updated by David van der Spoel about 9 years ago

Yes, I copied the files and reproduced the problem. Will try to fix it before release of 4.5.

#11 Updated by David van der Spoel about 9 years ago

This has been fixed for the upcoming 4.5 release.

Also available in: Atom PDF