Re: [Open Babel] openbabel renames hydrogens?

[Geoffrey H. <ge...@ge...>]

Sorry I didn't reply to your previous message -- I'm currently on  
vacation (before getting married).

First off, I think there was a typo in your 2nd message, which  
confused Noel:
> Well, here's a test pdb from insight2000 that contains only glycine  
> (note standard H names like HA1):

(And then a CAR file followed, then the PDB file from Insight.)

> I am not confusing file types and programs.  I have seen many, many
> molecular modeling and imaging programs utilize the .pdb file  
> format and
> different programs, in error perhaps, name atoms differently when
> exporting .pdb files.

Woah. Hold on a second. Yes, this is clearly an error in Open Babel's  
support of CAR, and we appreciate you reporting it. Honestly, I don't  
think our CAR code is as well-tested as we'd like because:
* None of the active developers use the format.
* There is no public, published standard for CAR that I can find.
* We don't have many authentic CAR files for testing.

Look, I like to keep things civil on this mailing list. We all want  
the same thing here -- to fix the bug you reported and help you  
convert files as you expect. One problem is that, as you say, many  
programs export PDB without necessarily sticking to the PDB spec.  
Open Babel tries very hard to stick to the published specification  
and atom names from the PDB itself.

Nevertheless, the problem isn't there in this case. (Although I see  
some strange PDB formatting, more on that in a second.)

 From the CAR code carformat.cpp:

>         // vs[0] contains atom label
>         // vs[4] contains "type of residue containing atom"
>         // vs[5] contains "residue sequence name"
>         // vs[6] contains "potential type of atom"

So the problem is that the carformat.cpp code ignores these columns  
altogether. So of course, the PDB output will have different names  
than your CAR input because it didn't read in the CAR names. That's  
one bug.

I'd personally really appreciate it if you could file a bug (and  
include the example files as attachments) to the bug database. The  
key problem with e-mail of text files is that mailing programs can do  
funny things with the formatting.
http://sourceforge.net/tracker/?atid=428740&group_id=40728&func=browse

That said, I think we can certainly fix this -- although if it's  
waiting for me to fix, it may be a few weeks until I'm back and have  
real network access.

Thanks and best regards,
-Geoff