comparison maxquant_mqpar.xml @ 3:2d67fb758956 draft

"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/maxquant commit da342a782ccc391b87fb4fead956b7b3cbd21258"

2:3fc2116ac6d9

<tool id="maxquant_mqpar" name="MaxQuant (using mqpar.xml)" version="@VERSION@">
    <macros>
        <import>macros.xml</import>
    </macros>
    <requirements>
        <requirement type="package" version="@VERSION@">maxquant</requirement>
        <requirement type="package" version="3.7">python</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
    #import re

    python3 '$__tool_directory__/mqwrapper.py'
    --num_threads=\${GALAXY_SLOTS:-1}
    --substitution_rx='@SUBSTITUTION_RX@'
    #if $infile.select == "mzxml_files"
        --mzxml_files='$infile.mzxml_files'
        #set names = ','.join([re.sub('@SUBSTITUTION_RX@', '_', str($name.element_identifier)) for $name in $infile.mzxml_files])
    #else
        --raw_files='$infile.raw_files'
        #set names = ','.join([re.sub('@SUBSTITUTION_RX@', '_', str($name.element_identifier)) for $name in $infile.raw_files])
    #end if
    --infile_names='$names'
    --version='@VERSION@'
    --fasta_files='$fasta_files'
    --identifier_parse_rule='$identifier_parse_rule'
    --description_parse_rule='$description_parse_rule'
    --mqpar_in='$mqpar_in'

    --evidence='$evidence'
    --msms='$msms'
    --parameters='$parameters'
    --peptides='$peptides'
    --proteinGroups='$proteinGroups'
    --allPeptides='$allPeptides'
    --libraryMatch='$libraryMatch'
    --matchedFeatures='$matchedFeatures'
    --modificationSpecificPeptides='$modificationSpecificPeptides'
    --ms3Scans='$ms3Scans'
    --msmsScans='$msmsScans'
    --mzRange='$mzRange'
    --peptideSection='$peptideSection'
    --summary='$summary'
    --mzTab='$mzTab'
    --output_all='$output_all'
    --mqpar_out='$mqpar_out'

    > '$log'
    ]]></command>
    <inputs>
        <conditional name="infile">
            <param name="select" type="select" label="choose the type of your input files">
                <option value="raw_files">Thermo.raw</option>
                <option value="mzxml_files">mzXML</option>
            </param>
            <when value="raw_files"> 
                <param multiple="true" name="raw_files" type="data"
                       format="thermo.raw" label="RAW Files"
                       help="Specify one or more Thermo RAW files" />
            </when>
            <when value="mzxml_files">
                <param multiple="true" name="mzxml_files" type="data"
                       format="mzxml" label="mzXML Files"
                       help="Specify one or more mzXML files" />
            </when>
        </conditional>
        <param format="fasta" multiple="true" name="fasta_files"
               type="data" label="FASTA files"
               help="Specify one or more FASTA databases." />

                <valid initial="string.printable">
                    <remove value="&apos;"/>
                </valid>
            </sanitizer>
        </param>
        <param type="data" name="mqpar_in" format="xml"
               label="mqpar.xml file with your search parameters. RAW file names 
                      must match the names displayed in galaxy. Their paths 
                      from the local machine are ignored. E.g. a file named
                      'test01.raw' in galaxy can either be named 'test01.raw' or
                      'D:\path\to\test01.raw' in the mqpar.xml."/>

        <param type="select" name="output" label="Select the desired outputs."
               multiple="true" optional="false">
            <expand macro="output_option" name="proteinGroups" label="Protein Groups"/>
            <expand macro="output_option" name="mqpar_out" label="mqpar.xml"/>
            <expand macro="output_option" name="peptides" label="Peptides"/>
            <expand macro="output_option" name="evidence" label="Evidence"/>
            <expand macro="output_option" name="parameters" label="Tabular Paramters"/>
            <expand macro="output_option" name="msms" label="MSMS"/>
            <expand macro="output_option" name="mzTab" label="mzTab"/>
            <expand macro="output_option" name="allPeptides" label="all peptides"/>
            <expand macro="output_option" name="libraryMatch" label="library match"/>
            <expand macro="output_option" name="matchedFeatures"
                    label="matched features"/>
            <expand macro="output_option" name="modificationSpecificPeptides"
                    label="modification specific peptides"/>
            <expand macro="output_option" name="ms3Scans" label="ms3 scans"/>
            <expand macro="output_option" name="msmsScans" label="msms scans"/>
            <expand macro="output_option" name="mzRange" label="mz range"/>
            <expand macro="output_option" name="peptideSection" label="peptide section"/>
            <expand macro="output_option" name="summary" label="summary"/>
            <expand macro="output_option" name="output_all"
                    label="complete 'combined/txt/' directory (compressed)"/>
        </param>
    </inputs>

    <outputs>
        <expand macro="output" name="mqpar_out" label="mqpar.xml" format="xml"/>
        <expand macro="output" name="proteinGroups" label="MaxQuant Protein Groups"/>
        <expand macro="output" name="peptides" label="MaxQuant Peptides"/>
        <expand macro="output" name="evidence" label="MaxQuant Evidence"/>
        <expand macro="output" name="parameters" label="MaxQuant Tabular Parameters"/>
        <expand macro="output" name="msms" label="MaxQuant MSMS"/>
        <expand macro="output" name="mzTab" label="mzTab"/>
        <expand macro="output" name="allPeptides" label="all peptides"/>
        <expand macro="output" name="libraryMatch" label="library match"/>
        <expand macro="output" name="matchedFeatures" label="matched features"/>
        <expand macro="output" name="modificationSpecificPeptides"
                label="modification specific peptides"/>
        <expand macro="output" name="ms3Scans" label="ms3 scans"/>
        <expand macro="output" name="msmsScans"
                label="msms Scans"/>
        <expand macro="output" name="mzRange"
                label="mz range"/>
        <expand macro="output" name="peptideSection"
                label="peptide section"/>
        <expand macro="output" name="summary"
                label="MaxQuant summary"/>
        <expand macro="output" format="tar" name="output_all"
                label="'combined/txt/' directory"/>
        <data name="log" format="txt"
              label="log"/>
    </outputs>

    <tests>
        <test>
            <param name="select" value="mzxml_files" />
            <param name="mzxml_files" value="BSA_min_22,BSA_min_21.mzXML" />
            <param name="fasta_files" value="bsa.fasta" />
            <param name="identifier_parse_rule" value="&gt;([^\s]*)" />
            <param name="description_parse_rule" value="&gt;(.*)" />
            <param name="mqpar_in" value="two/mqpar.xml" />
            <param name="output" value="evidence,msms,mzTab,allPeptides,msmsScans,mzRange,parameters,peptides,peptideSection,proteinGroups,summary,modificationSpecificPeptides" />
            <output name="evidence" file="two/combined/txt/evidence.txt" />
            <output name="msms" file="two/combined/txt/msms.txt" />
            <output name="mzTab" file="two/combined/txt/mzTab.mzTab" lines_diff="4"/>
            <output name="allPeptides" file="two/combined/txt/allPeptides.txt" lines_diff="14"/>
            <output name="msmsScans" file="two/combined/txt/msmsScans.txt" />
            <output name="mzRange" file="two/combined/txt/mzRange.txt" />
            <output name="parameters" file="two/combined/txt/parameters.txt" lines_diff="8"/>
            <output name="peptides" file="two/combined/txt/peptides.txt" />
            <output name="peptideSection" file="two/combined/txt/peptideSection.txt" />
            <output name="proteinGroups" file="two/combined/txt/proteinGroups.txt" />
            <output name="summary" file="two/combined/txt/summary.txt" />
            <output name="modificationSpecificPeptides" file="two/combined/txt/modificationSpecificPeptides.txt" />
        </test>
    </tests>
    <help><![CDATA[
MaxQuant is a quantitative proteomics software package designed for analyzing large mass-spectrometric data sets. 

This tool is a wrapper for MaxQuant v@VERSION@. It gets its search parameters from a previously created parameter file (mqpar.xml). A similiar tool that allows the specification of search parameters directly through galaxy is available as well and should be preferred. 

**Input files**

- Thermo raw file or mzXML file
    - The datatype has to be 'thermo.raw' or 'mzXML'. Make sure to specify the correct datatype either during upload to Galaxy or afterwards (edit attributes --> datatypes)

**Output files**

Different output file options are available, most of them are part of the MaxQuant txt folder. An additional mztab output option is implemented. 
    ]]></help>
    <citations>
        <citation type="bibtex">
        @article{cox2008maxquant,
        title={MaxQuant enables high peptide identification rates, individualized
        ppb-range mass accuracies and proteome-wide protein quantification},
        author={Cox, J{\"u}rgen and Mann, Matthias},
        journal={Nature biotechnology},
        volume={26},
        number={12},
        pages={1367},
        year={2008},
        publisher={Nature Publishing Group}
        }
        </citation>
        <citation type="bibtex">
        @article{tyanova2016maxquant,
        title={The MaxQuant computational platform for mass
        spectrometry-based shotgun proteomics},
        author={Tyanova, Stefka and Temu, Tikira and Cox, J{\"u}rgen},
        journal={Nature protocols},
        volume={11},
        number={12},
        pages={2301},
        year={2016},
        publisher={Nature Publishing Group}
        }
        </citation>
    </citations>
</tool>

3:2d67fb758956

<tool id="maxquant_mqpar" name="MaxQuant (using mqpar.xml)" version="@VERSION@">
    <macros>
        <xml name="output" token_format="tabular" token_label="default description" token_name="default">
            <data format="@FORMAT@" label="@LABEL@ for ${on_string}" name="@NAME@">
                <filter>'@NAME@' in output</filter>
            </data>
        </xml>
        <xml name="output_from_wdir" token_ext="txt" token_format="tabular" token_label="default description" token_name="default">
            <data format="@FORMAT@" from_work_dir="combined/txt/@NAME@.@EXT@" label="@LABEL@ for ${on_string}" name="@NAME@">
                <filter>'@NAME@' in output</filter>
            </data>
        </xml>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements"/>
    <command detect_errors="exit_code"><![CDATA[
    ## link galaxy datasets to filenames accepted by maxquant
    #import re
    #set names = [re.sub('@SUBSTITUTION_RX@', '_', str($n.element_identifier)) for $n in $input.infiles]
    #set names_with_ext = [($name if ($name).lower().endswith(str($input.ftype)) else $name + str($input.ftype)) for $name in $names]
    #for $target, $link in zip($infiles, $names_with_ext)
        ln -s '$target' '$link' &&
    #end for

    python3 '$__tool_directory__/modify_mqpar.py'
    #set inf = ','.join($names_with_ext)
    --infiles='$inf'
    --version=@VERSION@
    --num_threads=\${GALAXY_SLOTS:-1}
    --substitution_rx='@SUBSTITUTION_RX@'
    --fasta_files='$fasta_files'
    --description_parse_rule='$description_parse_rule'
    --identifier_parse_rule='$identifier_parse_rule'
    '$mqpar'

    &&
    maxquant mqpar.xml
    #if 'log' in $output:
        >> '$log'
    #end if
    && mv '$mqpar' combined/txt/mqpar.xml
    #if 'output_all' in $output:
        &&
        tar -zcf '$output_all' ./combined/txt
    #end if
  
    #if $qc.do_it:
        &&
        Rscript '$qr' '$qr_yaml'
        #if 'log' in $output:
            >> '$log' 2>&1
        #end if
        &&
        cp ./combined/txt/report_v0.92.6_combined.pdf '$ptxqc_report'
    #end if
    ]]></command>

    <configfiles>
        <expand macro="ptxqc"/>
    </configfiles>

    <inputs>
        <conditional name="input">
            <param name="ftype" type="select" label="choose the type of your input files">
                <option value=".thermo.raw">thermo.raw</option>
                <option value=".mzxml">mzXML</option>
                <option value=".mzml">mzML</option>
            </param>
            <when value=".thermo.raw"> 
                <param multiple="true" name="infiles" type="data"
                       format="thermo.raw" label="RAW Files"
                       help="Specify one or more Thermo RAW files" />
            </when>
            <when value=".mzxml">
                <param multiple="true" name="infiles" type="data"
                       format="mzXML" label="mzXML Files"
                       help="Specify one or more mzXML files" />
            </when>
            <when value=".mzml">
                <param multiple="true" name="infiles" type="data"
                       format="mzML" label="mzML Files"
                       help="Specify one or more mzML files" />
            </when>
        </conditional>
        <param format="fasta" multiple="true" name="fasta_files"
               type="data" label="FASTA files"
               help="Specify one or more FASTA databases." />

                <valid initial="string.printable">
                    <remove value="&apos;"/>
                </valid>
            </sanitizer>
        </param>
        <param type="data" name="mqpar" format="xml"
               label="mqpar.xml file with your search parameters. RAW file names 
                      must match the names displayed in galaxy. Their paths 
                      from the local machine are ignored. E.g. a file named
                      'test01.raw' in galaxy can either be named 'test01.raw' or
                      'D:\path\to\test01.raw' in the mqpar.xml."/>
        <expand macro="ptxqc-opts"/>
        <param type="select" name="output" label="Select the desired outputs."
               multiple="true" optional="false">
            <option value="proteinGroups">Protein Groups</option>
            <option value="mqpar">mqpar.xml</option>
            <option value="peptides">Peptides</option>
            <option value="evidence">Evidence</option>
            <option value="parameters">Tabular Paramters</option>
            <option value="msms">MSMS</option>
            <option value="mzTab">mzTab</option>
            <option value="allPeptides">all peptides</option>
            <option value="libraryMatch">library match</option>
            <option value="matchedFeatures">matched features</option>
            <option value="modificationSpecificPeptides">modification specific peptides</option>
            <option value="ms3Scans">ms3 scans</option>
            <option value="msmsScans">msms scans</option>
            <option value="mzRange">mz range</option>
            <option value="peptideSection">peptide section</option>
            <option value="summary">summary</option>
            <option value="output_all">complete 'combined/txt/' directory (compressed)</option>
            <option value="log">MaxQuant and PTXQC log</option>
        </param>
    </inputs>

    <expand macro="outputs"/>

    <tests>
        <test>
            <param name="ftype" value=".mzxml" />
            <param name="infiles" value="BSA_min_22,BSA_min_21.mzXML" />
            <param name="fasta_files" value="bsa.fasta" />
            <param name="identifier_parse_rule" value="&gt;([^\s]*)" />
            <param name="description_parse_rule" value="&gt;(.*)" />
            <param name="mqpar" value="mqpar.xml" />
            <param name="output" value="evidence,msms,mzTab,allPeptides,msmsScans,mzRange,parameters,peptides,peptideSection,proteinGroups,summary,modificationSpecificPeptides,output_all" />
            <output name="evidence" file="txt/evidence.txt" />
            <output name="msms" file="txt/msms.txt" lines_diff="2" />
            <output name="allPeptides" file="txt/allPeptides.txt" lines_diff="32" />
            <output name="msmsScans" file="txt/msmsScans.txt" lines_diff="4"/>
            <output name="mzRange" file="txt/mzRange.txt" />
            <output name="parameters" file="txt/parameters.txt" lines_diff="8"/>
            <output name="peptides" file="txt/peptides.txt" />
            <output name="proteinGroups" file="txt/proteinGroups.txt" />
            <output name="summary" file="txt/summary.txt" />
            <output name="modificationSpecificPeptides" file="txt/modificationSpecificPeptides.txt" />
        </test>
    </tests>
    <help><![CDATA[
MaxQuant is a quantitative proteomics software package designed for analyzing large mass-spectrometric data sets. 

This tool is a wrapper for MaxQuant v@VERSION@. It gets its search parameters from a previously created parameter file (mqpar.xml). A similiar tool that allows the specification of search parameters directly through galaxy is available as well and should be preferred, if possible.

**Input files**

- Thermo raw file or mzXML file
    - The datatype has to be 'thermo.raw' or 'mzXML'. Make sure to specify the correct datatype either during upload to Galaxy or afterwards (edit attributes --> datatypes)

**Output files**

Different output file options are available, most of them are part of the MaxQuant txt folder. An additional mztab output option is implemented. 
    ]]></help>
    <expand macro="citations"/>
</tool>