On 26/09/2011 11:50, Martin Guetlein wrote:
> On Mon, Sep 19, 2011 at 11:23 AM, Martin Guetlein<
> martin.guetl...@googlemail.com> wrote:
>
>> On Mon, Sep 19, 2011 at 11:04 AM, Chris Morley<c.mor...@gaseq.co.uk>
>> wrote:
>>> On 19/09/2011 09:32, Martin Guetlein wrote:
>>>> Hi,
>>>>
>>>> I would like to do smarts matching with OpenBabel, and for some
>>>> reasons I am restricted to use the command line interface.
>>>> Is there an efficient way to match a range of smarts strings against a
>> dataset?
>>>>
>>>> What I found out is that I could use the "-s" option, but I would have
>>>> to call babel once for each smarts, right?
>>> If the SMARTS patterns are sufficiently simple to be SMILES, then they
>>> can all be put in a file whose name is used with the -s option. The
>>> matching is OR - any single match will do. The file could be any any
>>> molecular format, but must have the appropriate extension.
>>
>
> I would have to run this once for each smarts, right? (This would produce
> quite a file parsing overhead.)
We are not quite understanding each other. Here is what I was meaning, I
hope it is what you want. There is a dataset (in any format) and you to
extract from it the molecules that match at least one of a set of
structures in a file patterns.yyy (any format). The output goes to
results.zzz You do a single command:
obabel dataset.xxx -O results.zzz -s patterns.yyy
The isomorphism filters are constructed once and applied for each input
molecule, but you don't have to worry about this.
>>>> A second option I found would be to edit the patterns.txt file and use
>>>> the FP3 fingerprint ("-ofpt -xfFP3 -xs"). I dont like the idea of
>>>> editing a file in advance, though (The smarts matching is part of some
>>>> software I am working on). Is there a way to provide an own patterns
>>>> file?
>>>
>>> You can make your own fingerprint by making a data file with SMARTS
>>> patterns, like patterns.txt (or with a couple of other slightly
>>> different formats) and specifying the details by making an entry in a
>>> plugindefines.txt. The MACCS fingerprint is defined there in this way.
>>
>> This looks good, I will give it a try, thanks,
>> Martin
>>
>
>
> Hmm, it worked fine in Linux, not for Windows, though. The plugindefines.txt
> is in a folder that requires admin rights to be edited (C:\Program
> Files\OpenBabel-2.3.0\data).
> Is there a way around that?
1) Copy the folder to somewhere accessible and set the environment
variable BABEL_DATADIR to the new position. Do this via Control
Panel/System and Security/System/Advanced system setting (in left
panel)/Environment Variables (button). You should see BABEL_DATADIR in
the top box.
or 2) Re-install OpenBabel to a different folder. (In the next version
the Windows Installer will use a different default location.)
>>
>>
>>>
>>> For extra flexibility you could define a compound filter to be used with
>>> the --filter option, where SMARTS tests can be combined with conditions
>>> based on other molecular (or SDF type) properties.
>>
>
> Same as for the -s option here, I have to run it once for each smarts,
> right?
You can run several SMARTS filters in the same command. Suppose you want
to only convert molecules that have a pyridine ring or have a bromine atom:
obabel dataset.xxx -O results.zzz --filter "s=n1ccccc1 || s=[Br]"
There seems to be a glitch with the compound filter, which would combine
these. I'll try to sort it out.
Chris
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
OpenBabel-discuss mailing list
OpenBabel-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/openbabel-discuss
