Pise/EMBOSS 2.3.1

Catherine Letondal letondal at pasteur.fr
Tue Apr 23 17:38:43 UTC 2002


Peter Rice wrote:
> Hi Catherine,

Hi Peter,


> 
> > Main changes were for align types, where I could associate a "pipetype" to
> > chain to other programs taking alignment as input. BTW, I found "MSF" and "fasta" for the
> > -aformat parameter - are there others?
> 
> There are more, but not sequence formats. We should add them to "entrails"
> output.
> 
> We can easily add more sequence formats. Can you suggest some?

I just asked to know which one to put on the Web interface.
(There are also clustalw or Phylip, but it's not necessary in Pise, since there
are format converters).
 

> Those with '*' are for pairwise alignments only.
> 
> > The main problem I had was with string parameters for specifying a path, with ./ default
> > value and having corresponding extn parameters.
> >
> > That's why I had to discard the following programs:
> > alignwrap, contacts, seqnr, seqsort, siggen, dichet and scopalign.
> >
> > A solution could be to have an explicit type and the extension as an attribute:
> > 
> > path: algpath  [
> >   parameter: "Y"
> >   prompt: "Location and extension of alignment files for input"
> >   default: "./"
> >   extn: ".align"
> > ]
> > 
> > What do you think?
> 
> The path and extension options  are a terrible 'hack' to avoid having "*"
> on the command line for those programs.

I have the same problem just with ./, since '/' cannot be allowed in a string
parameter on a Web server.

Another problem I have made a workaround for, is the '*' programs such as
extractseqfeat, where it is replaced in the Web form by 'all', then replaced
in the CGI by '*'. 

> 
> This is really just infile with a wild card filename (which works already).
> 
> We can make a new ACD type "inwild" which works like infile but with some
> small differences. The prompt would be "Input file(s)". The ajAcdGetInwild
> function will return an AjPFile. We can add functions to report the
> filenames as a string list (the first file is already open, the others are
> in a list so it is a little tricky to make the list in an application).
> 
> There should be an attribute "inextension:align" (for example) and a
> default value of "*". If the user specifies "*.align" the inextension will
> be ignored.
> 
> Associated qualifiers:
> 
> -inextension align
> -indirectory /home/user/somewhere (defaults to current directory)
> 
> For consistency, we can add the same qualifiers for infile.
> 
> With "out" instead of "in" we can sue the same qualifiers for outfile and a
> new ACD type "outwild" (outwild can open a new output file, using a new
> ajFileNextOut call, but the application needs to give the base name each
> time).
> 
> All easy to implement.
> 
> One problem ... inwild does not work well as a parameter because it has to
> be given as "*" on the command line. Same problem for "outwild". I am sure
> users can be educated.
> 
> The programs that use the path/extension options do not define them as
> parameters anyway. Their ACD files need some corrections.
> 
> Comments?

As long as there is a way to detect such kind of parameter (in order to replace
them by a simple textarea or file upload on a Web interface), I think it's
very useful! So the type would be inwild or outwild?

PS: Regarding Pise/EMBOSS I forgot to mention that not only output alignment 
are "connected" by Pise menus. I have also added this feature for sequence, seqall, 
seqout, etc...

Thanks for the quick answer!

--
Catherine Letondal -- Pasteur Institute Computing Center




More information about the EMBOSS mailing list