comparison protein_blast_grouping.xml @ 0:7abe5f471364 draft

planemo upload commit 7ebbd0df0aea9e58c4df58b61d6da385ee0ebb49

0:7abe5f471364

<tool id="edu.tamu.cpt.blast.protein_grouping" name="Protein Blast Grouping" version="0.0.1">
  <description>Based on a BLASTp result</description>
  <macros>
    <import>macros.xml</import>
  </macros>
  <expand macro="requirements"/>
  <command detect_errors="aggressive">
    <![CDATA[
      '$__tool_directory__/protein_blast_grouping.py'
      '${blast_in.blast}'
      --hits '$hits'
      --sort '$sort.sortType'
      > '$grouping_output'
    ]]>
  </command>
  <inputs>
    <conditional name="blast_in">
      <param name="blastType" type="select" label="Blastn Input Type">
        <option value="TSV">Blast Tabular</option>
      </param>
      <when value="TSV">
        <param label="BLASTp Results" name="blast" type="data" format="tsv,tabular"/>
      </when>
    </conditional>
    <param label="Number of results to return" name="hits" type="integer" value="5" min="1" max="30"/>
    <conditional name="sort">
      <param name="sortType" type="select" label="Sort by">
        <option value="unique_queries" selected="true">Unique Queries</option>
        <option value="unique_hits">Unique Hits</option>
      </param>
      <when value="unique_queries"/>
      <when value="unique_hits"/>
    </conditional>
  </inputs>
  <outputs>
    <data format="tabular" name="grouping_output" label="Top BlastP Hits"/>
  </outputs>
  <tests>
    <test>
      <conditional name="blast_in">
        <param name="blastType" value="TSV"/>
        <param name="blast" value="infile.txt"/>
      </conditional>
      <param name="hits" value="20"/>
      <output name="grouping_output" file="outfile.txt" lines_diff="1"/>
    </test>
  </tests>
  <help>
**What it does**
* Reads a tab-delimited BLAST output file.
* Extracts organism names from the subject titles (text in square brackets).
* Counts unique query proteins that matched each organism and unique hit proteins from each organism.
* Sorts and displays results based on either unique queries or unique hits.
* The output is a formatted table showing the top N organisms with the most matches.
</help>
  <expand macro="citations-2020"/>
</tool>