[Biopython] BioPython wrapper for UNAFOLD and NUPACK

[Rouilly, Vincent]

Hi,

many thanks for your feedbacks about UNAFOLD.

I completely agree with the fact that one has to be careful with the Perl Script packaging involved in UNAFOLD.
As suggested, I'll get in touch with their development team to check if they have any intention to provide python support.
At the same time, within the next week, I'll work on providing more documentation on API + Perl script functions to this list.

And I'll do the same for NUPACK. In that case, it should be simpler as there are only binaries involved.

thanks again for your inputs,
best,
Vincent.


________________________________________
From: p.j.a.cock at googlemail.com [p.j.a.cock at googlemail.com] On Behalf Of Peter [biopython at maubp.freeserve.co.uk]
Sent: Thursday, June 18, 2009 10:30 AM
To: Michiel de Hoon
Cc: Rouilly, Vincent; biopython at lists.open-bio.org
Subject: Re: [Biopython] BioPython wrapper for UNAFOLD and NUPACK

On Thu, Jun 18, 2009 at 2:19 AM, Michiel de Hoon <mjldehoon at yahoo.com> wrote:
>
> I'm a bit biased here, since I use UNAFold a lot for my own research.
>
> One thing to keep in mind is that UNAFold relies a lot on Perl scripts that
> glue the actual executables together. A Biopython interface can either run
> the Perl scripts (which would introduce a Perl dependency), or replicate
> the Perl scripts in Python (which is more difficult to maintain, but may give
> us a more Pythonic way to run UNAFold). You could also consider to
> contact the UNAFold developers directly; they may be interested in a
> Python wrapper in addition to the Perl wrapper to their software (so, the
> Python wrapper would be part of UNAFold rather than of Biopython).

If UNAFold is a collection of Perl scripts which call some compiled code,
then the natural thing would just be to wrap the Perl scripts just like any
other command line tool. I presume they see the Perl scripts as the
public API.

UNAFold isn't the only command line tool to use Perl internally, for
example the main SignalP executable is also a Perl script. Many of
these tools will be Unix/Linux only where Perl is normally installed
anyway - I don't see this indirect Perl dependency as a problem.
i.e. If you want to use UNAFold, you need Perl. If you want to call
UNFold from Biopython, you need UNAFold, therefore you also need
Perl. This would be an optional runtime dependency like any other
command line tool we wrap. This doesn't mean Biopython needs Perl ;)

If the underlying compiled code could be wrapped directly in Python
that may be more elegant, but does really require input from UNAFold
themselves. It would be worth investigating.

Peter