[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: linking apps in Debian-Med with data



Le Tue, Feb 19, 2008 at 09:57:10AM +0100, Steffen Moeller a écrit :
> 
> you may have become aware of the trunk/community/infrastructure/getData
> effort to download databases and to spawn the one or other application
> once the download has completed. It works, in a perl hash you
>   ...
>   "identify the database" => {
> 	name => "describe it a bit more",
> 	source => "say how to mirror it and ",
> 	"post-download" => "list what to do once the files have appeared locally"
>   },
>   ...
> Easy, isn't it?
> 
> At the moment I am using that post-download feature only to unzip the
> prior downloaded files - indexing for blast or so is probably a better
> textbook-like use case but not mine, really :o/   Is there anybody out
> there who would like to have a look at getData to add some more
> sophisticated dependencies to the Debian-(Med|Science) infrastructure?

Hi Steffen,

I have commited change to make easier to build the getdata Debian
package. I then have tried to see how to index a fasta database
(swiss.fasta for instance) for emboss after download. The operation is
not so complex, but encoding shell commands in a perl hash is kind of
painful. I was considering putting this in a separate file,
/usr/lib/getdata/post-install/emboss-index for instance, but it still is
a bit uncomfortable, as we need to manipulate long file paths to give
proper arguments to that kind of script. I am thinking about using a ref
to a perl subroutine instead of a quoted shell script. Would it make
sense to you? Also, it might be possible to use BioPerl to call
EMBOSS indexing facilities.

PS: getData is probably too generic: if we want to convince our
upstreams not to use such names, we have to show the example!

Have a nice day,

-- 
Charles


Reply to: