Hi all,

An eagle eyed user has just spotted a bug in our BLAST wrappers
and/or the Galaxy framework itself.

1. Run BLASTN (wrapper version v0.1.00 which is currently on the Test Tool Shed)
2. Select BLASTXML output, execute
3. While running click the "i" icon, which wrongly reports:

Format:tabular

4. Check the "repeat", which correctly has XML output
5. Once finished, notice the output is XML, but Galaxy labelled it "tabular"
6. Workaround via the "pencil" icon, correct the datatype to "blastxml".

This happens both on our production server:

$ hg summary
parent: 11218:26f58e05aa10 release_2013.11.04
 Galaxy stable release for 2013.11.04.
branch: stable

And also on my development server with galaxy-central:

$ hg summary
parent: 13675:bc1173309bd5 tip
 Variable name fix for the tool shed's install and test framework.
branch: default

This used to work perfectly. We've not changed this area of the code
in the BLAST wrappers directly - it uses a default output type of "tabular"
and the <change_format> tag - expanding the macro we have:

    <outputs>
        <data name="output1" format="tabular"
label="${blast_type.value} $query.name vs @ON_DB_SUBJECT@">
            <!--  <expand macro="output_change_format" /> -->
            <change_format>
                <when input="out_format" value="0" format="txt"/>
                <when input="out_format" value="0 -html" format="html"/>
                <when input="out_format" value="2" format="txt"/>
                <when input="out_format" value="2 -html" format="html"/>
                <when input="out_format" value="4" format="txt"/>
                <when input="out_format" value="4 -html" format="html"/>
                <when input="out_format" value="5" format="blastxml"/>
            </change_format>
        </data>
    </outputs>

What has changed recently in the NCBI BLAST wrappers is
that input parameter "out_format" is now nested within a
conditional block.

Other tools using <change_format> seem to still work fine:
https://github.com/peterjc/pico_galaxy/tree/master/tools/seq_filter_by_id

Can anyone testing the BLAST+ update on the Test Tool Shed
reproduce this? Any thoughts on the cause and fix? I have to
attend a meeting now, so I've not dug any deeper yet...

Regards,

Peter
___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:
  http://lists.bx.psu.edu/

To search Galaxy mailing lists use the unified search at:
  http://galaxyproject.org/search/mailinglists/

Reply via email to