[MMTK] dcd_to_nc : IOError: Database entry Atoms/ not found

[Konrad Hinsen]

On Aug 3, 2008, at 5:34, Harindar Keer wrote:

> Many thanks for providing the detailed explanation, it got me into  
> thinking in terms of tracking the source of the problems. The  
> format of pdb file was incorrect and also there are more than 100,  
> 000 atoms numbered as ***** after 99999, that was also creating the  
> trouble. After changing every such "*****" entry to 99999 it works  
> and no further errors are reported  but I am skeptical if there  
> would be problem in further analysis of the system.

The atom numbers are not used anywhere by MMTK, so your workaround  
should not cause any trouble.

> Now in some cases, I have been running into DCD read error. For the  
> same system it works for some dcd files but not on others, I am  
> still not able to narrow down to specific cause of the problem.
>
> Traceback (most recent call last):
>  File "/usr/bin/dcd_to_nc", line 121, in <module>
>    dcd_reader()
>  File "/usr/lib/python2.5/site-packages/MMTK/DCD.py", line 64, in  
> __call__
>    self.getActions(), self.getOption('dcd_file'))
> IOError: DCD read error

If this occurs very early in the conversion (before you have any  
simulation steps in the netCDF file), this probably indicates a  
format problem. MMTK uses a rather old DCD reader which cannot read  
some more recent variants such as DCD files containing variable box  
shape data at each time step. I will send you a test version of a  
newer DCD->netCDF converter that knows about those newer formats, but  
hasn't been tested much yet.

If the error occurs when some steps have already been converted, your  
DCD has probably been corrupted. You could check if you can read it  
completely with other tools, e.g. VMD.

Konrad.
--
---------------------------------------------------------------------
Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: hinsen at cnrs-orleans.fr
Web: http://dirac.cnrs-orleans.fr/~hinsen/
---------------------------------------------------------------------