[vtk-developers] caffeine.tcl regression

[Kyle Lutz]

The PDB format is notoriously hard to implement as many programs and
scripts output their own dialect of the format.

This article has some good tips on robustly parsing PDB files:
http://www.daylight.com/meetings/mug01/Sayle/m4xbondage.html

-kyle

On Tue, Oct 16, 2012 at 4:34 AM, Bill Lorensen <bill.lorensen at gmail.com> wrote:
> Joachim,
>
> I agree with your analysis. That caffeine file is not a valid PDB file. I
> think your patch is a good one. I tried in on several valid PDB files and it
> worked great.
>
> Thanks,
>
> Bill
>
> On Tue, Oct 16, 2012 at 3:58 AM, Joachim Pouderoux
> <joachim.pouderoux at kitware.com> wrote:
>>
>> Hi Bill,
>>
>> I am not an expert of this domain at all but here is what I understood of
>> the situation.
>> I also suspect the caffeine file not being a valid PDB file. According the
>> latest 2011 file format specification but also the 1998 one, the element
>> symbol of an atom should be specified at columns 77-78, while the caffeine
>> file does not provide it. The previous PDB reader used to take the 2 first
>> chars of the name field (columns 13-16) to get the atom symbol (through the
>> method vtkMoleculeReaderBase::MakeAtomType()) while this field use a very
>> specific nomenclature - "described in the Chemical Component Dictionary" and
>> this caused a misinterpretation of most PDB files I fetched on the PDB
>> database.
>> So to get the atom symbol with "official" PDB files and with this old and
>> may be singular caffeine.pdb file, the latest patch I committed try to use
>> the element field first and use name only of the element field is empty.
>>
>> Regards,
>> Joachim
>>
>>
>> 2012/10/14 Bill Lorensen <bill.lorensen at gmail.com>
>>>
>>> Ooops. I wrote this e-mail before I saw that you had updated the gerrit
>>> topic: http://review.source.kitware.com/#/t/1451/
>>>
>>> I just tried your patch and caffeine.tcl produces the correct image.
>>>
>>> Thanks,
>>>
>>> Bill
>>>
>>>
>>> On Sun, Oct 14, 2012 at 8:40 AM, Bill Lorensen <bill.lorensen at gmail.com>
>>> wrote:
>>>>
>>>> Joachim,
>>>>
>>>> I noticed that you have a gerrit topic for the caffeine.tcl regression.
>>>>
>>>> It seems that the file VTKData/Data/caffeine.pdb is not a valid pdb
>>>> file. The file has been there since 2003, but I suspect it did not come from
>>>> the official PDB. It may have come from here:
>>>> http://elchem.kaist.ac.kr/jhkwak/okanaganpdb97/molecule/molecule.html
>>>> The author says "I use a commercial programme to obtain x,y,z
>>>> coordinates and use these in a format usable by RasMol."
>>>>
>>>> 1) I suggest we add the missing element symbols to caffeine.pdb.
>>>> 2) Since the new PDB code ignores hydrogen (I assume they is some
>>>> chemistry reason for this), I suggest we add a new baseline that shows
>>>> caffeine without hydrogen.
>>>>
>>>> With these 2 changes, there is no need to modify the current PDB reader.
>>>>
>>>> Bill
>>>>
>>>
>>>
>>>
>>> --
>>> Unpaid intern in BillsBasement at noware dot com
>>>
>>
>
>
>
> --
> Unpaid intern in BillsBasement at noware dot com
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Follow this link to subscribe/unsubscribe:
> http://www.vtk.org/mailman/listinfo/vtk-developers
>
>