Bug #2052
trjconv does not recognize periodic molecules
Description
Running trjconv on a periodic molecule yields
There were 4 inconsistent shifts. Check your topology
despite the tpr file knowing about this.
This obviously never worked and therefore this is a feature request rather than a bug.
Related issues
History
#1 Updated by Berk Hess over 4 years ago
- Tracker changed from Feature to Bug
- Affected version set to 4.6
This is a bug. The pbc is read from the tpr and you should not be able to make a graph with periodic molecules. This used to work, but I see now that the (new, but by now old) rmpbc.cpp file never gets the inputrec passed.
Note that the only thing we need to add is a fatal error. This is trivial to do in gmx_trjconv.cpp. But it would be better to do it in rmpbc, because otherwise other tools might have the same issue.
#2 Updated by David van der Spoel over 4 years ago
I would prefer not to have a fatal error since then I can not process my trajectories anymore. Will look into a real fix. But should a solution really go into 4.6? I guess the latest supported version is 5.1 so that should be sufficient, right?
#3 Updated by Berk Hess over 4 years ago
I don't understand your last comment. Calling rmpbc on a periodic molecule should result in a fatal error. The "solution" is not to ask for molecules to be made whole. Or are there cases where you get useful output from an operation that includes rmpbc?
Note that I set the affected version to 4.6. I left the target at 2016.
#4 Updated by David van der Spoel over 4 years ago
I have a complex system with a periodic DNA molecule and non-periodic proteins. The proteins need to be made whole but not the DNA... Maybe I need to do a couple of trjconv runs in this case.
#5 Updated by Berk Hess over 4 years ago
That part is indeed a feature request. A simple solution would be to not complain about inconsistent shifts, potentially only when the periodic molecule option is set.
#6 Updated by Mark Abraham about 4 years ago
- Target version deleted (
2016.1)
Please set target versions when someone plans to do something concrete, rather than have some possible fiction that gets bumped with every patch release :-)
#7 Updated by Mark Abraham almost 4 years ago
- Related to Task #2115: trjconv does too many things, and combinations of them work poorly added
#8 Updated by Gerrit Code Review Bot about 3 years ago
Gerrit received a related patchset '1' for Issue #2052.
Uploader: Paul Bauer (paul.bauer.q@gmail.com)
Change-Id: gromacs~master~Ief60ad46191ea17a7cbc7d181bf1727195ff9677
Gerrit URL: https://gerrit.gromacs.org/7464