Re: [Open Babel] (no subject)

[Richard G. <re...@co...>]

Geoff Hutchison wrote:

> On Wednesday, January 22, 2003, at 09:30  AM, Richard Gillilan wrote:
>
>> OBResidueData::AssignBonds is used to add bonds and bond orders and
>> is called during ReadPDB. It creates peptide (inter-residue) bonds
>> between residues using a distance criterion alone and then tries
>> to recognize the residue type and assign intra-residue bonds from
>> the appropriate entry in the resdata.txt file. It then rejects
>> any intra-residue bonds longer than 3.5 Angstroms.
>
>
> I killed this for other reasons. Unfortunately people do use 
> PDB-format non-protein files and this was creating C-N bonds in very 
> strange places. ConnectTheDots() seems to do a better job and uses 
> IMHO more chemical cutoffs.
>
> That said, I'd like to know if you've had a chance to check the 
> current PDB and Mol2 code. It should keep residue information when 
> converting Mol2->PDB, but I don't use either format enough to be sure 
> I haven't done something strange.

I am happily using OB to assign bond orders and hydrogens to proteins 
(from PDB) and
writing out as mol2 for use in my docking code. I have not had time to 
test anything
beyond proteins and will continue to be in production mode for several 
more weeks.
After that, I am anxious to run the MMFF94 test structures through OB to 
see how
well it does in general.

(I did recently have to add a leading space in the write pdb part of the 
code. The treatment
  of leading spaces (Calcium vs. CA etc.) is tricky in PDB ... I saw 
there were some recent
additions to that section of code, so perhaps the problem is solved in a 
more general
fashion than my quick fix

>
> In particular, I'm concerned about atoms with names like "HG11" that 
> are hydrogens... Should they become "1HG1" in the PDB column to make 
> sure the 'H' character ends up in the correct place? The PDB 2.2 
> standard seems to indicate that this should be resolved with the 
> element in columns 77-78, but how do older files and codes do this?

I prefer to see a leading space here and then just step on the alternate 
location indicator column for the
last digit. Alternate locations are not used very much and must usually 
be discarded anyway when you
set up a structure for simulation (you have to choose one conformation). 
I know of very few programs
if any that read the alt indicator and do something useful with it. It 
is mainly output from crystallographers
who see multiple conformations of sidechains.  My own code does not care 
about
unique atom names anyway.

Richard

>
> -Geoff
>
>