[Bioperl-l] Re: Bio::Tools::Run prepare executions [was Re:bioperl-run Codeml.pm fix_blength]

[Jason Stajich]

On Aug 3, 2005, at 11:00 AM, Albert Vilella wrote:

> Hi all,
>
> Having thought about the previous thread on changing tempdir as a
> settable value in Bio::Tools::Run::WrapperBase (Jason? should we?)...
>
i think it will be fine to do those changes if I remember correctly  
what they were... =)
> ...I wonder if it may be interesting (at least it would for me) to
> have something like a "prepare" method for the execution wrappers in
> Bio::Tools::Run.
>
> What I'm looking for is a way to create the dirs corresponding to the
> analysis one wants to conduct. The "prepare" method would create, but
> not execute, the dir with the ready-to-run elements of the executables
> according to the various input data files and parameters.
>
> Right now, we have a "run" method that first prepares the elements
> needed for the execution and then runs the program.
>
> We also have container objects for program results in bioperl-live.
>
> This "prepare" method might be useful for people wanting to generate
> sets of analysis for further execution on queueing-based systems or
> similar scheduled execution situations.
>

Sure - this sounds fine- I guess part of the prepare step, though is  
preparing the arguments to send to the programs.
Do you want to capture these arguments as well?
My understanding is the BioPipe system (which may not have many devs  
now) tried make this possible by encoding the input options to the  
Perl modules in an XML file which was loaded into the pipeline db.
http://www.genome.org/cgi/content/abstract/13/8/1904

But I'm definitely open to some other ideas about how this should be  
done and the idea of a prepare step seems great (especially if we  
break out a cleanup step as well and insure that every run cmd does a  
prepare, execute, cleanup cycle.

Thanks for jumping in on this - I think your ideas and intuition here  
are right on the mark and I think a more systematic approach on the  
parts needed to run an external program should be spelled out in the  
code.

-jason
> I agree that the sole "preparation" of an execution it might not fit
> well with the idea of an execution wrapper as it is now in bioperl, so
> any suggestions/comments/criticism are welcome.
>
> Bests,
>
>     Albert.
>
>
>
> El dt 12 de 07 del 2005 a les 18:03 +0200, en/na Albert Vilella va
> escriure:
>
>> El dt 12 de 07 del 2005 a les 11:47 -0400, en/na Jason Stajich va
>> escriure:
>>
>>> Sounds good - would you just copy the dir to the users specified
>>> outdir?
>>>
>>
>> yes
>>
>>
>>>    Another way to go is make tempdir a settable value (see
>>> Bio::Tools::Run::WrapperBase -- in bioperl-live repository) - but
>>> this may not be as clear on how to use it?
>>>
>>
>> well, it is not as direct as the other way but maybe it is cleaner in
>> the sense that will directly run the analysis on $tempdir and no  
>> extra
>> cp or mv would be needed...
>>
>>    Albert.
>>
>>
>>>
>>>
>>> -jason
>>> On Jul 12, 2005, at 11:40 AM, Albert Vilella wrote:
>>>
>>>
>>>> El dt 12 de 07 del 2005 a les 11:28 -0400, en/na Jason Stajich va
>>>> escriure:
>>>>
>>>>
>>>>> sure - fix away.
>>>>>
>>>>>
>>>>
>>>>
>>>> done.
>>>>
>>>>
>>>> Also, in my pipeline it would be interesting to call Codeml.pm via
>>>> bioperl keeping the tempfiles in a specified directory:
>>>>
>>>>
>>>> I understand that save_tempfiles will save the generated tempfiles
>>>> in
>>>> the temp directory, the dir will remain in $tempdir.
>>>> An $outdir could be specified so that the codeml run is saved where
>>>> the
>>>> user specifies.
>>>>
>>>>
>>>> What do you think?
>>>>
>>>>
>>>>     Albert.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>> --
>>>
>>> Jason Stajich
>>>
>>> jason.stajich at duke.edu
>>>
>>> http://www.duke.edu/~jes12/
>>>
>>>
>>>
>>>
>>>
>>>
>
>

--
Jason Stajich
Duke University
http://www.duke.edu/~jes12