<tool id="secimtools_add_kegg_pathway_info" name="Add KEGG Pathway Information" version="@WRAPPER_VERSION@">
  <description>using KEGGIDs</description>
  <macros>
    <import>macros.xml</import>
  </macros>
  <expand macro="requirements" />
  <command detect_errors="exit_code"><![CDATA[
    add_kegg_pathway_info.py
      -sp=$species
      #if $dataSets.whichDataSet == "geneDataset":
        -gka=$dataSets.geneKeggAnnot
        -gid=$dataSets.geneUniqId
        -gn=$dataSets.geneName
        -gkid=$dataSets.geneKeggId
        -go=$geneOutput
        -kg2p=$gene_keggID2pathID
      #end if
      #if $dataSets.whichDataSet == "metDataset":
        -mka=$dataSets.metKeggAnnot
        -mid=$dataSets.metUniqId
        -mn=$dataSets.metName
        -mkid=$dataSets.metKeggId
        -mo=$metOutput
        -km2p=$met_keggID2pathID
      #end if
      #if $dataSets.whichDataSet == "both":
        -gka=$dataSets.geneKeggAnnot
        -gid=$dataSets.geneUniqId
        -gn=$dataSets.geneName
        -gkid=$dataSets.geneKeggId
        -mka=$dataSets.metKeggAnnot
        -mid=$dataSets.metUniqId
        -mn=$dataSets.metName
        -mkid=$dataSets.metKeggId
        -kg2p=$gene_keggID2pathID
        -km2p=$met_keggID2pathID
        -go=$geneOutput
        -mo=$metOutput
      #end if
      -p=$pathways
  ]]></command>
  <inputs>
    <param name="species" type="select" label="Select your Species from the List" >
     <option value="hsa">Homo sapiens</option>
     <option value="mmu">Mus musculus</option>
     <option value="rno">Rattus norvegicus</option>
     <option value="dme">Drosophila melanogaster</option>
     <option value="ath">Arabidopsis thaliana</option>
     <option value="sce">Saccharomyces cerevisiae</option>
     <option value="eco">Escherichia coli</option>
     <option value="cel">Caenorhabditis elegans</option>
    </param>
    <conditional name="dataSets">
      <param name="whichDataSet" type="select" display="radio" label="Select the Datasets you want to add KEGG pathway information to.  Note: datasets must contain KEGGIDs.">
        <option value="both" selected="true">Gene Expression + Metabolomic Files with KEGGIDs</option>
        <option value="geneDataset">Gene Expression File with KEGGIDs</option>
        <option value="metDataset">Metabolomic File with KEGGIDs</option>
        <validator type="no_options" message="You must select at least one option." />
      </param>
      <when value="both">
        <param name="geneKeggAnnot" type="data" format="tabular" label="Gene to KEGGID Link File" help="Select the Gene to KEGGID Link File that contains KEGG identifiers from your history"/>
        <param name="geneUniqId" type="text" size="30" value="" label="Gene unique FeatureID" help="Name of the column in your Gene to KEGGID Link File that contains unique FeatureIDs."/>
        <param name="geneName" type="text" size="30" value="" label="Gene Symbol" help="Name of the column in your Gene to KEGGID Link File that contains Gene Symbols."/>
        <param name="geneKeggId" type="text" size="30" value="" label="Gene KEGGID" help="Name of the column in your Gene to KEGGID Link File that contains KEGGIDs."/>
        <param name="metKeggAnnot" type="data" format="tabular" label="Metabolite to KEGGID Link File" help="Select the Metabolite to KEGGID Link File that contains KEGG identifiers from your history"/>
        <param name="metUniqId" type="text" size="30" value="" label="Metabolite unique FeatureID" help="Name of the column in your Metabolite to KEGGID Link File that contains unique FeatureIDs."/>
        <param name="metName" type="text" size="30" value="" label="Metabolite Names" help="Name of the column in your Metabolite to KEGGID Link File that contains Metabolite Names."/>
        <param name="metKeggId" type="text" size="30" value="" label="Metabolite KEGGID" help="Name of the column in your Metabolite to KEGGID Link File that contains KEGGIDs."/>
      </when>
      <when value="geneDataset">
        <param name="geneKeggAnnot" type="data" format="tabular" label="Gene to KEGGID Link File" help="Select the Gene to KEGGID Link File that contains KEGG identifiers from your history"/>
        <param name="geneUniqId" type="text" size="30" value="" label="Gene unique FeatureID" help="Name of the column in your Gene to KEGGID Link File that contains unique FeatureIDs."/>
        <param name="geneName" type="text" size="30" value="" label="Gene Symbol" help="Name of the column in your Gene to KEGGID Link File that contains Gene Symbols."/>
        <param name="geneKeggId" type="text" size="30" value="" label="Gene KEGGID" help="Name of the column in your Gene to KEGGID Link File that contains KEGGIDs."/>
      </when>
      <when value="metDataset">
        <param name="metKeggAnnot" type="data" format="tabular" label="Metabolite to KEGGID Link File" help="Select the Metabolite to KEGGID Link File that contains KEGG identifiers from your history"/>
        <param name="metUniqId" type="text" size="30" value="" label="Metabolite unique FeatureID" help="Name of the column in your Metabolite to KEGGID Link File that contains unique FeatureIDs."/>
        <param name="metName" type="text" size="30" value="" label="Metabolite Names" help="Name of the column in your Metabolite to KEGGID Link File that contains Metabolite Names."/>
        <param name="metKeggId" type="text" size="30" value="" label="Metabolite KEGGID" help="Name of the column in your Metabolite to KEGGID Link File that contains KEGGIDs."/>
      </when>
    </conditional>
  </inputs>
  <outputs>
    <data format="tabular" name="gene_keggID2pathID" label="${tool.name} on ${on_string}: GeneKeggID2PathwayID">
      <filter>(dataSets['whichDataSet'] == 'geneDataset') or (dataSets['whichDataSet'] == 'both')</filter>
    </data>
    <data format="tabular" name="met_keggID2pathID" label="${tool.name} on ${on_string}: MetaboliteKeggID2PathwayID">
      <filter>(dataSets['whichDataSet'] == 'metDataset') or (dataSets['whichDataSet'] == 'both')</filter>
    </data>
    <data format="tabular" name="pathways" label="${tool.name} on ${on_string}: PathwayID2PathwayNames"/>
    <data format="tabular" name="geneOutput" label="${tool.name} on ${on_string}: Gene KEGG Pathway File">
      <filter>(dataSets['whichDataSet'] == 'geneDataset') or (dataSets['whichDataSet'] == 'both')</filter>
    </data>
    <data format="tabular" name="metOutput" label="${tool.name} on ${on_string}: Metabolite KEGG Pathway File">
      <filter>(dataSets['whichDataSet'] == 'metDataset') or (dataSets['whichDataSet'] == 'both')</filter>
    </data>
  </outputs>
  <tests>
    <test>
      <param name="species" value="rno"/>
      <param name="whichDataSet" value="both"/>
      <param name="geneKeggAnnot" value="gene_to_keggId_link.tsv"/>
      <param name="geneUniqId" value="UniqueID"/>
      <param name="geneName" value="GeneSymbol"/>
      <param name="geneKeggId" value="KEGG_ID"/>
      <param name="metKeggAnnot" value="metabolite_to_keggId_link.tsv"/>
      <param name="metUniqId" value="UniqueID"/>
      <param name="metName" value="MetName"/>
      <param name="metKeggId" value="KEGG_ID"/>
      <output name="gene_keggID2pathID" file="geneKeggId2pathwayId.tsv" compare="diff" lines_diff="10000"/>
      <output name="met_keggID2pathID" file="metaboliteKeggId2pathwayId.tsv" compare="diff" lines_diff="10000"/>
      <output name="geneOutput" file="gene_kegg_pathway.tsv" compare="diff" lines_diff="10000"/>
      <output name="metOutput" file="metabolite_kegg_pathway.tsv" compare="diff" lines_diff="10000"/>
      <output name="pathways" file="pathwayId2pathwayNames.tsv" compare="diff" lines_diff="10000"/>
    </test>
  </tests>
  <help><![CDATA[

**Tool Description**

  This tool takes a Gene to KEGG Link dataset or a Metabolomic to KEGG Link dataset and adds KEGG Pathway Names using KEGGIDs.
  The tool was designed to take the output from the 'Link Name to KEGGID' tool as input (for example the Gene to KEGGID Link dataset),
  but other datasets containing KEGGIDs can be used as well.


  If a 'Gene to KEGG ID link File’ is given, the tool outputs the following three files;

	1) a 'Gene KEGG Pathway File' containing the unique FeatureID, Feature_Name, Feature_Type and KEGG_ID columns from the input file plus Pathway_ID and Pathway_Name columns 	containing KEGG pathway identifiers and KEGG pathway names, respectively.  If no pathways are 	linked to the KEGG IDs then these columns return ‘NA’.

	2) a 'GeneKeggID2PathwayID' file containing the KEGG gene identifiers and their associated KEGG pathway identifiers.

	3) a 'PathwayID2PathwayNames' file containing all of the KEGG pathway identifiers and theirassociated KEGG pathway names.

  If a 'Metabolite to KEGG ID link File’ is given, the tool outputs the following three files;

	1) a ' Metabolite KEGG Pathway File' containing the unique FeatureID, Feature_Name, Feature_Type and KEGG_ID columns from the input file plus Pathway_ID and Pathway_Name 	columns containing KEGG pathway identifiers and KEGG pathway names, respectively.  If no 	pathways are linked to the KEGG IDs then these columns return ‘NA’.

	2) a 'MetaboliteKeggID2PathwayID' file containing the KEGG metabolite identifiers and their associated KEGG pathway identifiers.

	3) a 'PathwayID2PathwayNames' file containing all of the KEGG pathway identifiers and their associated KEGG pathway names.


  Note: FeatureIDs and KEGGIDs may not be unique in the output.

--------------------------------------------------------------------------------

**INPUT**

**Gene Dataset containing KEGGIDs - for example, output from the Link Name to KEGG IDtool**

  +----------+-----------+--------------+---------+--------------+---------+------------+-----+----------+
  | UniqueId | Gene_Name | Feature_Type | Matched | Name_in_KEGG | KEGGID  | Similarity | Tie | Selected |
  +==========+===========+==============+=========+==============+=========+============+=====+==========+
  | Gene_1   | one       | Gene         | Yes     | one*         | mmu:... | 1.0        | No  | Yes      |
  +----------+-----------+--------------+---------+--------------+---------+------------+-----+----------+
  | Gene_2   | two       | Gene         | Yes     | two*         | mmu:... | 1.0        | No  | Yes      |
  +----------+-----------+--------------+---------+--------------+---------+------------+-----+----------+
  | Gene_3   | three     | Gene         | Yes     | three*       | mmu:... | 1.0        | No  | Yes      |
  +----------+-----------+--------------+---------+--------------+---------+------------+-----+----------+
  | Gene_4   | four      | Gene         | No      | NA           | NA      | NA         | No  | NA       |
  +----------+-----------+--------------+---------+--------------+---------+------------+-----+----------+
  | ...      | ...       | ...          | ...     | ...          | ...     | ...        | ... | ...      |
  +----------+-----------+--------------+---------+--------------+---------+------------+-----+----------+

**Metabolite Dataset containing KEGGIDs - for example, output from the Link Name to KEGGID tool**

  +----------+----------+--------------+---------+--------------+----------+------------+-----+----------+
  | UniqueId | Met_Name | Feature_Type | Matched | Name_in_KEGG | KEGGID   | Similarity | Tie | Selected |
  +==========+==========+==============+=========+==============+==========+============+=====+==========+
  | Met_1    | one      | Metabolite   | Yes     | one*         | cpd:...  | 1.0        | No  | Yes      |
  +----------+----------+--------------+---------+--------------+----------+------------+-----+----------+
  | Met_2    | two      | Metabolite   | Yes     | two*         | cpd:...  | 1.0        | No  | Yes      |
  +----------+----------+--------------+---------+--------------+----------+------------+-----+----------+
  | Met_3    | three    | Metabolite   | Yes     | three*       | cpd:...  | 1.0        | No  | Yes      |
  +----------+----------+--------------+---------+--------------+----------+------------+-----+----------+
  | Met_4    | four     | Metabolite   | No      | NA           | NA       | NA         | No  | NA       |
  +----------+----------+--------------+---------+--------------+----------+------------+-----+----------+
  | ...      | ...      | ...          | ...     | ...          | ...      | ...        | ... | ...      |
  +----------+----------+--------------+---------+--------------+----------+------------+-----+----------+

**Gene unique FeatureID and/or Metabolite unique ID**

  Name of the column in your KEGGID Link File that contains unique identifiers (for genes and/or metabolites).

**Gene Symbol**

  Name of the column in your KEGGID Link File that has Gene Symbols (for genes) or Metabolite Names (for metabolite data).

**Gene KEGGID and/or Metabolite KEGGID**

  Name of the column in your KEGGID Link File that contains KEGGIDs (for genes and/or metabolites).

--------------------------------------------------------------------------------

**OUTPUT**

  The user will get different output from the tool, depending on whether they include the 'Gene to KEGGID Link' File, the 'Metabolite to KEGGID Link' File, or both.

  (1) **GeneKeggID2PathwayID.** Downloaded file from KEGG for the selected species that contains ALL KEGGIDs to PathwayIDs.
  (2) **MetaboliteKeggID2PathwayID.** Downloaded file from KEGG for the selected species that contains ALL the Metabolite KeggIDs to PathwayIDs.
  (3) **Pathways Names.** Downloaded file from KEGG for the selected species that contains ALL the PathwayIDs to Pathway Names.
  (4) **Gene KEGG Pathway File.** Tabular file with genes, feature types, KEGGIDs, PathwayIDs and Pathway Names.

  +-------------+--------------+--------------+---------+--------------+--------------+
  | FeatureID   | Gene_Symbol  | Feature_Type | KEGGID  | PathwayID    | Pathway_Name |
  +=============+==============+==============+=========+==============+==============+
  | FeatureID_1 | one          | Gene         | mmu:... | path:mmu:... | Pathway_A    |
  +-------------+--------------+--------------+---------+--------------+--------------+
  | FeatureID_2 | two          | Gene         | mmu:... | path:mmu:... | Pathway_B    |
  +-------------+--------------+--------------+---------+--------------+--------------+
  | FeatureID_3 | three        | Gene         | mmu:... | path:mmu:... | Pathway_C    |
  +-------------+--------------+--------------+---------+--------------+--------------+
  | FeatureID_4 | four         | Gene         | mmu:... | path:mmu:... | Pathway_D    |
  +-------------+--------------+--------------+---------+--------------+--------------+
  | ...         | ...          | ...          | ...     | ...          | ...          |
  +-------------+--------------+--------------+---------+--------------+--------------+

------------------------------------------------------------------

  (5) **Metabolite KEGG Pathway File.** Tabular file with metabolites, feature types, KEGGIDs, PathwayIDs and Pathway Names.

  +-------------+-----------------+--------------+---------+--------------+--------------+
  | FeatureID   | Metabolite_Name | Feature_Type | KEGGID  | PathwayID    | Pathway_Name |
  +=============+=================+==============+=========+==============+==============+
  | FeatureID_1 | one             | Metabolite   | cdp:... | path:map:... | Pathway_A    |
  +-------------+-----------------+--------------+---------+--------------+--------------+
  | FeatureID_2 | two             | Metabolite   | cdp:... | path:map:... | Pathway_B    |
  +-------------+-----------------+--------------+---------+--------------+--------------+
  | FeatureID_3 | three           | Metabolite   | cdp:... | path:map:... | Pathway_C    |
  +-------------+-----------------+--------------+---------+--------------+--------------+
  | FeatureID_4 | four            | Metabolite   | cdp:... | path:map:... | Pathway_D    |
  +-------------+-----------------+--------------+---------+--------------+--------------+
  | ...         | ...             | ...          | ...     | ...          | ...          |
  +-------------+-----------------+--------------+---------+--------------+--------------+

  ]]>
  </help>
  <citations>
    <citation type="bibtex">@ARTICLE{Kirpich17secimtools,
    author = {Alexander S. Kirpich, Miguel Ibarra, Oleksandr Moskalenko, Justin M. Fear, Joseph Gerken, Xinlei Mi, Ali Ashrafi, Alison M. Morse, Lauren M. McIntyre},
    title = {SECIMTools: A suite of Metabolomics Data Analysis Tools},
    journal = {BMC Bioinformatics},
    year = {2018}
    }</citation>
    <citation type="bibtex">@article{Mor2021GaitGM,
    title={GAIT-GM integrative cross-omics analyses reveal cholinergic defects in a C. elegans model of Parkinson's disease},
    author={Mor, DE and Huertas, F and Morse, AM and Kaletsky, R and Murphy, CT and Kalia, V and Miller, GW and Moskalenko, O and Conesa, A and McIntyre, LM},
    journal={BMC Genomics},
    year={submitted},
    }</citation>
  </citations>
</tool>