<tool id="edu.tamu.cpt.external.psm_recombine" name="PSM Prep and Table Generator:" version="21.0.0">
  <description>: generates a raw PSM file for plotting and a tabular summary</description>
  <macros>
    <import>macros.xml</import>

  </macros>
  <requirements>
    <requirement type="package" version="5.22.1">perl</requirement>
    <requirement type="package">perl-bioperl</requirement>
    <requirement type="package">perl-moose</requirement>
    <requirement type="package">perl-ipc-run</requirement>
    <requirement type="package">perl-getopt-long-descriptive</requirement>
    <requirement type="package" version="3.7">python</requirement>
    <requirement type="package" version="1.74">biopython</requirement>
  </requirements>
  <command detect_errors="aggressive"><![CDATA[
perl -Mlib='$__tool_directory__/lib' '$__tool_directory__/cpt_psm_0_prep.pl'
--galaxy
--outfile_supporting '$__new_file_path__'
#for $item in $repeat_file:
--file "${item.file}"
#end for

--cpt_psm_object "${cpt_psm_object}"

--cpt_psm_object_files_path "${cpt_psm_object.files_path}"

--cpt_psm_object_format "TXT"

--cpt_psm_object_id "${cpt_psm_object.id}" 

&&

mkdir tmp_001
&&
mkdir tmp_002

&&

perl -Mlib='$__tool_directory__/lib' '$__tool_directory__/cpt_psm_2_gentable.pl'
--galaxy
--outfile_supporting '$__new_file_path__'
--file "${cpt_psm_object}"

#if $evalue and $evalue is not "None":
--evalue "${evalue}"
#end if

#if $dice and $dice is not "None":
--dice "${dice}"
#end if

#if $mismatch and $mismatch is not "None":
--mismatch "${mismatch}"
#end if

#if $gap_penalty and $gap_penalty is not "None":
--gap_penalty "${gap_penalty}"
#end if

#if $match and $match is not "None":
--match "${match}"
#end if

--diff_table "./tmpDiff.tsv"
--diff_table_files_path "./tmp_001"
--diff_table_format TSV_U
--diff_table_id "000001"

--blastclust "./tmpClust.tsv"
--blastclust_files_path "./tmp_002"
--blastclust_format TSV_U
--blastclust_id "000002"

&&

'$__tool_directory__/PSM_Recombine.py'
./tmpDiff.tsv
#for $item in $repeat_file:
"${item.file}"
#end for
> "${tableOut}" 
]]>
</command>
  <inputs>
    <repeat name="repeat_file" title="Genbank file(s) used to generate PSM">
      <param name="file" optional="False" label="Gbk file" type="data" format="genbank"/>
    </repeat>
    <param value="0.0001" name="evalue" help="Evalue cutoff" optional="True" label="evalue" type="float"/>
    <param value="50" name="dice" help="Dice cutoff" optional="True" label="dice" type="float"/>
    <param value="-1" name="mismatch" help="Mismatch Score" optional="True" label="mismatch" type="float"/>
    <param value="0.0" name="gap_penalty" help="Gap Penalty" optional="True" label="gap_penalty" type="float"/>
    <param value="5" name="match" help="Match Score" optional="True" label="match" type="float"/>
  </inputs>
  <outputs>
    <data format="tabular" name="tableOut" label="PSM Tabular Results"/>
    <data name="cpt_psm_object" label="Raw PSM Binary output" format="TXT"/>
  </outputs>
  <help><![CDATA[
**What it does**

This tool takes in 2 or more GenBank files, blasts, and prepares data
structures for use in the companion tool, PSM Plotter, as well as a 
tabular file read-out of the results. Select as many (multi)-gbk files 
as you *might* want to plot. Once this tool is done, you can select 
any subset of those to plot then.

IMPORTANT PARAMETERS
--------------------

-  ``mismatch``, ``gap_penalty``, ``match``

   These parameters control the Needleman-Wunsch Multiple Sequence
   Alignment library's scoring scheme. Mismatch scores are generally
   negative and discourage unrelated proteins from being plotted in a
   line together. Match scores encourage related proteins to line up.
   Gap penalty is set at zero as we generally prefer gaps to mismatches
   in this tool; phage genomes are small and gaps are "cheap" to use,
   whereas mismatches can sometimes give an incorrect impression of
   relatedness. That said, how your plots look is completely up to you
   and we encourage experimentation!
  ]]></help>
  <expand macro="citations-2020"/>
</tool>