view mothur/tools/mothur/get.sharedseqs.xml @ 25:bfbaf823be4c

Change metagenomics datatypes to include labels and groups metadata. change Mothur tool configs to get label and group select options from a data_meta filter rather than using the options from_dataset attribute. This grealty decreases memory demand for the galaxy server.

<tool id="mothur_get_sharedseqs" name="Get.sharedseqs" version="1.20.0" force_history_refresh="True">
 <description>Get shared sequences at each distance from list and group</description>
 <command interpreter="python">
  mothur_wrapper.py 
  #import re, os.path
  --cmd='get.sharedseqs'
  --result='^mothur.\S+\.logfile$:'$logfile
  --outputdir='$logfile.extra_files_path'
  --list=$list
  --group=$group
  #set datasets = ["'^\S+?\.((unique|[0-9.]*)(\S+)\.shared.seqs)$:tabular'"]
  #if $label.__str__ != "None" and len($label.__str__) > 0:
   --label='$label'
  #end if
  #if $seqs_from.selection == 'unique':
   #if $seqs_from.groups.__str__ != "None" and len($seqs_from.groups.__str__) > 0:
    --unique=$seqs_from.groups
   #end if
  #elif $seqs_from.selection == 'shared':
   #if $seqs_from.groups.__str__ != "None" and len($seqs_from.groups.__str__) > 0:
    --shared=$seqs_from.groups
   #end if
  #end if
  #if $fasta.__str__ != "None" and len($fasta.__str__) > 0:
   --fasta=$fasta
   #set datasets = $datasets + ["'^\S+?\.((unique|[0-9.]*)(\S+)\.shared.fasta)$:" + $fasta.ext + "'"]
  #end if
  #if $output.__str__ != "None" and len($output.__str__) > 0:
   --output=$output
   #set datasets = $datasets + ["'^\S+?\.((unique|[0-9.]*)(\S+)\.accnos)$:accnos'"]
  #end if
  #if $as_datasets.__str__ == "yes":
   --datasetid='$logfile.id' --new_file_path='$__new_file_path__' 
   --new_datasets=#echo ','.join($datasets)
  #end if
 </command>
 <inputs>
  <param name="list" type="data" format="list" label="list - OTU List"/>
  <param name="group" type="data" format="groups" label="group - "/>
  <param name="label" type="select" optional="true" label="label - Select OTU Labels to include" multiple="true" 
     help="By default all are included if no selection is made.">
   <options>
    <filter type="data_meta" ref="list" key="labels" />
   </options>
  </param>
  <param name="fasta" type="data" format="fasta" optional="true" label="fasta - Dataset"/>

  <conditional name="seqs_from">
   <param name="selection" type="select" label="Select Groups" help="">
    <option value="all" selected="true">None (use all groups)</option>
    <option value="unique">OTUs that contain ONLY sequences from the selected groups</option>
    <option value="shared">OTUs that contain sequences from the selected groups</option>
   </param>
   <when value="all"/>
   <when value="unique">
    <param name="groups" type="select" label="unique - Group to analyze" multiple="true">
     <options>
      <filter type="data_meta" ref="group" key="groups" />
     </options>
    </param>
   </when>
   <when value="shared">
    <param name="groups" type="select" label="shared - Groups to analyze" multiple="true">
     <options>
      <filter type="data_meta" ref="group" key="groups" />
     </options>
    </param>
   </when>
  </conditional> <!-- seqs_from -->
  
  <param name="output" type="select" label="output - select the output format" help="accnos can be used with get.seqs, list.seqs and remove.seqs">
    <option value="" selected="true">default: (name  group  bin_number)</option>
    <option value="accnos">accnos (name)</option>
  </param>
  <param name="as_datasets" type="boolean" truevalue="yes" falsevalue="no" checked="true" label="Create a new history dataset for each label"/>
 </inputs>
 <outputs>
  <data format="html" name="logfile" label="${tool.name} on ${on_string}: logfile" />
 </outputs>
 <requirements>
  <requirement type="binary">mothur</requirement>
 </requirements>
 <tests>
 </tests>
 <help>
**Mothur Overview**

Mothur_, initiated by Dr. Patrick Schloss and his software development team
in the Department of Microbiology and Immunology at The University of Michigan,
provides bioinformatics for the microbial ecology community.

.. _Mothur: http://www.mothur.org/wiki/Main_Page

**Command Documenation**

The get.sharedseqs_ command takes a list and group file and outputs a .shared.seqs file for each distance. This is useful for those cases where you might be interested in identifying sequences that are either unique or shared by specific groups, which you could then classify.

.. _get.sharedseqs: http://www.mothur.org/wiki/Get.sharedseqs

 </help>
</tool>