<tool id="checkm_tree_qa" name="CheckM tree_qa" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
    <description>
        Assess phylogenetic markers in the genome tree
    </description>
    <macros>
        <import>macros.xml</import>
        <xml name="concatenated_tre">
            <param name="concatenated_tre" type="data" format="phyloxml" label="Concatenated tree"/>
        </xml>
    </macros>
    <expand macro="biotools"/>
    <expand macro="requirements"/>
    <expand macro="version"/>
    <command detect_errors="exit_code"><![CDATA[
@HMM_MARKER_STATS_INPUTS@
mkdir 'inputs/storage/tree/' &&
#if $output.out_format != '5'
ln -s '$output.concatenated_tre' 'inputs/storage/tree/concatenated.tre' &&
#else
ln -s '$output.concatenated_fasta' 'inputs/storage/tree/concatenated.fasta' &&
#end if
 
checkm tree_qa
    'inputs'
    --out_format $output.out_format
    --tab_table
    --file 'output_file'
    ]]></command>
    <inputs>
        <expand macro="hmm_marker_stats_inputs"/>
        <conditional name="output">
            <param argument="--out_format" type="select" label="Desired output">
                <option value="1">Brief summary of genome tree placement</option>
                <option value="2">Detailed summary of genome tree placement including lineage-specific statistics</option>
                <option value="3">Genome tree in Newick format decorated with IMG genome ids</option>
                <option value="4">Genome tree in Newick format decorated with taxonomy strings</option>
                <option value="5">Multiple sequence alignment of reference genomes and bins</option>
            </param>
            <when value="1">
                <expand macro="concatenated_tre"/>
            </when>
            <when value="2">
                <expand macro="concatenated_tre"/>
            </when>
            <when value="3">
                <expand macro="concatenated_tre"/>
            </when>
            <when value="4">
                <expand macro="concatenated_tre"/>
            </when>
            <when value="5">
                <param name="concatenated_fasta" type="data" format="phyloxml" label="Concatenated masked sequences"/>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_f1" format="tabular" from_work_dir="output_file" label="${tool.name} on ${on_string}: Summary of genome tree placement">
            <filter>output['out_format']=="1"</filter>
        </data>
        <data name="output_f2" format="tabular" from_work_dir="output_file" label="${tool.name} on ${on_string}: Summary of genome tree placement including lineage-specific statistics">
            <filter>output['out_format']=="2"</filter>
        </data>
        <data name="output_f3" format="newick" from_work_dir="output_file" label="${tool.name} on ${on_string}: Genome tree decorated with IMG genome ids">
            <filter>output['out_format']=="3"</filter>
        </data>
        <data name="output_f4" format="newick" from_work_dir="output_file" label="${tool.name} on ${on_string}: Genome tree decorated with taxonomy strings">
            <filter>output['out_format']=="4"</filter>
        </data>
        <data name="output_f5" format="tabular" from_work_dir="output_file" label="${tool.name} on ${on_string}: Multiple sequence alignment of reference genomes and bins">
            <filter>output['out_format']=="5"</filter>
        </data>
    </outputs>
    <tests>
        <test expect_num_outputs="1">
            <param name="phylo_hmm_info" ftype="zip" value="phylo_hmm_info.pkl.gz"/>
            <param name="bin_stats_tree" ftype="tabular" value="bin_stats.tree.tsv"/>
            <param name="hmmer_tree">
                <collection type="list">
                    <element name="637000110" ftype="txt" value="hmmer.tree.txt"/>
                </collection>
            </param>
            <conditional name="output">
                <param name="out_format" value="1"/>
                <param name="concatenated_tre" ftype="phyloxml" value="concatenated.tre"/>
            </conditional>
            <output name="output_f1" ftype="tabular">
                <assert_contents>
                    <has_text text="637000110"/>
                    <has_text text="k__Bacteria;p__Proteobacteria;c__Gammaproteobacteria"/>
                    <has_text text="# unique markers"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="phylo_hmm_info" ftype="zip" value="phylo_hmm_info.pkl.gz"/>
            <param name="bin_stats_tree" ftype="tabular" value="bin_stats.tree.tsv"/>
            <param name="hmmer_tree">
                <collection type="list">
                    <element name="637000110" ftype="txt" value="hmmer.tree.txt"/>
                </collection>
            </param>
            <conditional name="output">
                <param name="out_format" value="2"/>
                <param name="concatenated_tre" ftype="phyloxml" value="concatenated.tre"/>
            </conditional>
            <output name="output_f2" ftype="tabular">
                <assert_contents>
                    <has_text text="637000110"/>
                    <has_text text="# unique markers (of 43)"/>
                    <has_text text="UID5220"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="phylo_hmm_info" ftype="zip" value="phylo_hmm_info.pkl.gz"/>
            <param name="bin_stats_tree" ftype="tabular" value="bin_stats.tree.tsv"/>
            <param name="hmmer_tree">
                <collection type="list">
                    <element name="IMG_646564547" ftype="txt" value="hmmer.tree.txt"/>
                </collection>
            </param>
            <conditional name="output">
                <param name="out_format" value="3"/>
                <param name="concatenated_tre" ftype="phyloxml" value="concatenated.tre"/>
            </conditional>
            <output name="output_f3" ftype="newick">
                <assert_contents>
                    <has_text text="IMG_646564547"/>
                    <has_text text="UID7|f__Methanocaldococcaceae"/>
                    <has_text text="f__Archaeoglobaceae"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="phylo_hmm_info" ftype="zip" value="phylo_hmm_info.pkl.gz"/>
            <param name="bin_stats_tree" ftype="tabular" value="bin_stats.tree.tsv"/>
            <param name="hmmer_tree">
                <collection type="list">
                    <element name="637000110" ftype="txt" value="hmmer.tree.txt"/>
                </collection>
            </param>
            <conditional name="output">
                <param name="out_format" value="4"/>
                <param name="concatenated_tre" ftype="phyloxml" value="concatenated.tre"/>
            </conditional>
            <output name="output_f4" ftype="newick">
                <assert_contents>
                    <has_text text="IMG_646564547"/>
                    <has_text text="k__Archaea;p__Euryarchaeota"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="phylo_hmm_info" ftype="zip" value="phylo_hmm_info.pkl.gz"/>
            <param name="bin_stats_tree" ftype="tabular" value="bin_stats.tree.tsv"/>
            <param name="hmmer_tree">
                <collection type="list">
                    <element name="637000110" ftype="txt" value="hmmer.tree.txt"/>
                </collection>
            </param>
            <conditional name="output">
                <param name="out_format" value="5"/>
                <param name="concatenated_fasta" ftype="fasta" value="concatenated.fasta"/>
            </conditional>
            <output name="output_f5" ftype="tabular">
                <assert_contents>
                    <has_text text=">637000110"/>
                    <has_text text="-TVNQLVR"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
@HELP_HEADER@

This command assesses phylogenetic markers in the genome tree

Inputs
======


Outputs
=======

Output in function of selection output format

1. brief summary of genome tree placement indicating the number of unique phylogenetically informative markers found, the number of markers found multiple times, and a taxon string indicating the placement of each bin within the genome tree
2. detailed summary of genome tree placement giving a more detailed indication of where each bin is within the genome tree, general characteristics about each bin (e.g., GC, genome size, coding density), and general characteristics about all reference genomes descendant from the parental node of each bin (e.g., mean and standard deviation of GC)
3. genome tree in Newick format decorated with IMG genome ids which can be used to examine the phylogenetic neighbours of each bin
4. genome tree in Newick format decorated with taxonomy strings which can be used to examine the phylogenetic neighbours of each bin
5. multiple sequence alignment of reference genomes and bins which can be used to infer a de novo genome tree


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