Feature #902

Suggestion for better pdb-compliance

Added by Erik Marklund over 8 years ago. Updated over 6 years ago.

analysis tools
Target version:


Several parts of gromacs (mdrun, analysis tools) sometimes break the pdb format when a number, e.g., a coordinate or a b-factor, is too large for the allowed precision. This may render the pdb file unreadable by gromacs tools and other programs, which is quite problematic. What is the best way to deal with it? Putting in a number that fits, but is not the right one, like 999.999; or breaking the format, which makes the file unreadable? In principle one could do both: capping the numbers so that they adhere to the format, and adding a REMARK that contains the correct number. The pdb-reading functions in gromacs could make use of this extra information. Whenever writing or reading such data a notification/warning should be sent to e.g. stderr.


#1 Updated by Mark Abraham over 8 years ago

This issue has probably been addressed before. We should make some attempt to follow others' solutions.

#2 Updated by Erik Marklund over 8 years ago

MacroMolecular Builder uses REMARKs to add coordinates with more precision.

#3 Updated by Roland Schulz about 8 years ago

@Erik. Your suggested solution doesn't seem to be to complicated. Could you upload a patch for it?

#4 Updated by Erik Marklund about 8 years ago

I could. Unfortunately, it's application season for me, so it'll have to wait until the end of the month or so.

#5 Updated by Erik Marklund about 8 years ago

  • Assignee changed from David van der Spoel to Erik Marklund

Should I have 4.6 as the target version?

#6 Updated by Erik Lindahl over 6 years ago

  • Tracker changed from Bug to Feature
  • Status changed from New to Closed

Nothing has happened after a patch was offered two years, and it's not something of high priority to other developers, so I'm rejecting/closing this. Feel free to reopen when a patch is ready.

For the record, to strictly follow the PDB format all REMARK lines need to appear before any ATOM line, which would require a bit of more work. This is incidentally the same reason why we haven't added velocities in PDB files the same way.

#7 Updated by Erik Marklund over 6 years ago

I think this was the wrong route to follow anyway. I've opened up a related feature [[]]

Also available in: Atom PDF