Bug #163

"g_bond -d" reports long skewed bond that shouldn't be there

Added by Sean Holbert over 13 years ago. Updated over 13 years ago.

Erik Lindahl
analysis tools
Target version:
Affected version - extra info:
Affected version:


I'm using GROMACS version 3.1.4. When I use "g_bond -d" tool, it reports that one of the bonds is 13nm long in the first frame and gets longer and longer as the simulation progresses. By the last frame of the simulation, the bond is ~16nm long. Attached are the trajectory over 10 picoseconds of the simulation, the topology, and index for the protein, so you can run the tool and see the skewed bonds for yourself. The bond occurs at the 88th residue, between a carbon and hydrogen.

test (39 Bytes) test Data needed to run "g_bond -d" tool. The tool however reports a huge bond at residue 88. Sean Holbert, 09/13/2007 10:42 PM


#1 Updated by Sean Holbert over 13 years ago

Created an attachment (id=234)
Data needed to run "g_bond -d" tool. The tool however reports a huge bond at residue 88.

#2 Updated by Erik Lindahl over 13 years ago

Please see if you can repeat this with 3.3.1, or (much better) the release-3-3-branch of CVS since we're about to release 3.3.2. Unfortunately we won't fix the bug if it's only in 3.1.4 :-)

#3 Updated by David van der Spoel over 13 years ago

I think you have misinterpreted the manual. According to g_bond -h:

g_bond makes a distribution of bond lengths. If all is well a gaussian
distribution should be made when using a harmonic potential. bonds are read
from a single group in the index file in order i1-j1 i2-j2 thru in-jn.

This mean you can not simply give it the protein index file as it will interpret your list of atoms as bonds, some of them maybe quite far off.

Also available in: Atom PDF