Re: [Open Babel] Problem using fs index file

[Chris M. <c.m...@ds...>]

Omar Haq wrote:
> Yes, the original structures file is in the same directory.
> new_ligands.sdf is it's name.
> 
> I used >babel new_ligands.sdf -ofs
> to make the index.
> 
> I don't believe I get an error while making the index. This is the
> output after making the index:
>> babel new_ligands.sdf -ofs
> This will prepare an index of new_ligands.sdf and may take some time...
>  It took 49 seconds
> 8365 molecules converted
> 15 audit log messages
> 
> This is what happens after I did the -osdf to read the headers:
> 
>> babel new_ligands.fs -osdf
> ==============================
> *** Open Babel Error  in ReadChemObject
>  Difficulty reading from index new_ligands.fs
> 
> 0 molecules converted
> 1 errors 1 audit log messages
> 
> 
> When I make the index from a short list of smiles, my index file is
> actually empty...Is that normal? I believe when I was installing
> openBabel on my linux machine, it told me about something needed to be
> updated. Something related to xml..

The index file should be about 36 bytes for each molecule + about 256 
bytes for the header, so it sounds as if the index is not being made 
properly.

Is your system 64 bit or does it somehow use default wide characters? 
The binary operations during indexing may not have been written robustly 
enough to handle these, and have not been tested on such systems.

If you look at the index file in a binary editor, can you see the name 
of the original data file near the beginning?

> Actually the reason I am doing all of this, is because I actually want
> to compare ever single structure in new_ligands.sdf to itself, and not
> jus the first structure on the list. Is there a quick way I can do
> that, or would  you recommend I write code to wrap around babel so I
> can make an N by N matrix of the tanimoto scores for my compounds?

OpenBabel's capabilities are not a good match to this problem, at the 
moment. I suspect 'external' scripting which calls the OB command 
interface would be too slow. 'Internal' scripting using Python, Perl or 
Ruby which calls the fingerprint routines directly would be better. See, 
for instance, 
http://baoilleach.blogspot.com/2007/07/pybel-just-how-unique-are-your_12.html
But my feeling is that these n squared tasks are best done in the C++ 
code - if it existed.

Coincidentally this morning, Chris Swain sent me a reference to a 
clustering program
http://cactus.nci.nih.gov/SUBSET/
It sounds as if you may be doing something related. I'm looking into 
whether the functionality of this program could be incorporated into 
OpenBabel.

Chris
> 
> Thanks!
> 
> On 8/23/07, Chris Morley <c.m...@ds...> wrote:
>> Omar Haq wrote:
>>> After creating the fs file, babel doesn't seem to read it. I was
>>> wondering if anyone could guide me as to what the reason might be.
>>> This is the error I am getting:
>>>
>>>> babel new_ligands.fs results.sdf -Starget -at0.8
>>> ==============================
>>> *** Open Babel Error  in ReadChemObject
>>>   Difficulty reading from index new_ligands.fs
>>>
>>> 0 molecules converted
>>> 1 errors 1 audit log messages
>>>
>> Is the original datafile, maybe called new_ligands.sdf, present in the
>> same directory as new_ligands.fs? The fs file is just an index and needs
>> read data from the original file.
>>
>> Could you show the commandline when you made the index?
>>
>> Try
>> babel new_ligands.fs -osdf
>> This should give information on the fs file, if it can read its header.
>>
>> Do you always get the same problem, even when making an index of a file
>> with only a small set of molecules?
>>
>> Chris
>>
>>
>>
>>
>> -------------------------------------------------------------------------
>> This SF.net email is sponsored by: Splunk Inc.
>> Still grepping through log files to find problems?  Stop.
>> Now Search log events and configuration files using AJAX and a browser.
>> Download your FREE copy of Splunk now >>  http://get.splunk.com/
>> _______________________________________________
>> OpenBabel-discuss mailing list
>> Ope...@li...
>> https://lists.sourceforge.net/lists/listinfo/openbabel-discuss
>>
> 
>