<tool id="picrust_categorize" name="Categorize" version="@TOOL_VERSION@.0">
    <description>by collapsing hierarchical data to a specified functional level</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="bio_tools"/>
    <expand macro="requirements"/>
    <expand macro="version_command"/>
    <command detect_errors="aggressive"><![CDATA[
        categorize_by_function.py
            -i '$input_fp'
            -c $metadata_category
            -l $level
            -o tempbiom

            @OUTPUT_CONVERSION_COMMANDS@
    ]]></command>
    <inputs>
        <param argument="--input_fp" format="biom1,h5" type="data" label="Input file" help="Biom file that is output from the PICRISt predict tool"/>
        <param argument="--metadata_category" type="select" multiple="false" label="Metadata category that describes the hierarchy" help="">
            <option value="KEGG_Pathways" selected="true">KEGG Pathways</option>
            <option value="COG_Category">COG Category</option>
            <option value="taxonomy">Using taxonomy metadata</option>
        </param>
        <param argument="--level" type="integer" value="3" min="1" label="Hierarchy Level"
               help="a value of 1 is the highest level, and any higher value nears the leaves of the hierarchy. For instance, if the hierarchy contains 4 levels, specifying 3 would collapse at one level above being fully specified"/>
        <expand macro="biom_format_select"/>
    </inputs>
    <outputs>
        <expand macro="biom_output"/>
    </outputs>
    <tests>
        <test> <!-- test with Kegg Pathways and json output -->
            <param name="input_fp" value="predicted_metagenomes.L3.biom"/>
            <param name="metadata_category" value="KEGG_Pathways"/>
            <param name="level" value="2"/>
            <param name="output_type" value="json"/>
            <output name="out_biom" ftype="biom1">
                <assert_contents>
                    <has_text text="Biological Observation Matrix"/>
                    <has_text text="generated_by"/>
                    <has_text text="KEGG_Pathways"/>
                    <has_text text="Cancers"/>
                    <not_has_text text="ABC transporters"/>
                </assert_contents>
            </output>
        </test>
        <test> <!-- test with COG categories and classic output -->
            <param name="input_fp" value="cog_predicted_metagenomes.L2.biom"/>
            <param name="metadata_category" value="COG_Category"/>
            <param name="level" value="1"/>
            <param name="output_type" value="tsv"/>
            <output name="out_biom" ftype="tabular" md5="7682875d365f884b399ffa521952dbd8"/>
        </test>
        <test> <!-- test with taxonomy metadata and hdf5 output -->
            <param name="input_fp" value="observation_table.biom"/>
            <param name="metadata_category" value="taxonomy"/>
            <param name="level" value="1"/>
            <param name="output_type" value="hdf5"/>
            <output name="out_biom" ftype="h5" file="categorize_biom.hdf5" compare="sim_size"/>
        </test>
    </tests>
    <help>
<![CDATA[
@PICRUST_OVERVIEW@

**Command Documentation**

This module collapses hierarchical data to a specified level for PICRUSt predictions. For instance, often it is useful to examine KEGG results from a higher level within the pathway hierarchy.
Many genes are sometimes involved in multiple pathways, and in these circumstances (also know as a one-to-many relationship), the gene is counted for each pathway.

**Input file:** Output file from the predict tool

]]>
    </help>
    <expand macro="citations"/>
</tool>