<tool id="quast" name="Quast" version="4.1" >
    <description>Genome assembly Quality</description>
    <requirements>
        <requirement type="package" version="4.1">quast</requirement>
    </requirements>
    <stdio>
        <regex match="ERROR! exception caught!"
            source="both"
            level="fatal"
            description="Something went wrong. Check the following errors : No labels..." />
    </stdio>
    <command>
    <![CDATA[
        quast
        -o outputdir
        #if $gene_selection == "eukaryote":
            --eukaryote
        #else if $gene_selection == "metagenes":
            --meta
        #end if
        #if $input_ref:
            -R '$input_ref'
            #if $input_operon:
                -O '$input_operon'
            #end if
            #if $annot:
                -G '$annot'
            #end if
        #end if
        #if $input_size:
            --est-ref-size $input_size
        #end if
        --min-contig $min_contig
        --contig-thresholds $threshold_contig
        #for $k in $files:
            #if $k.type_file == "scaffold":
                -s
            #end if
        $k.input
        #end for
        &&
        mv outputdir/report.txt '$quast_out' &&
        mv outputdir/report.tsv '$quast_tsv' &&
        mv outputdir/report.tex '$quast_tex' &&
        mv outputdir/icarus_viewers/contig_size_viewer.html '$icarus' &&
        mv outputdir/report.html '$report_html' 
    ]]>
    </command>
    <inputs>
        <repeat name="files" title="Input assembly files" min="1">
            <param type="data" format="fasta" name="input" label="Contigs/scaffolds output file"/>
            <param name="type_file" type="select" label="Type of data">
                <option value="contig">Contig</option>
                <option value="scaffold">Scaffold</option>
            </param>
        </repeat>
        <param name="input_size" type="integer" label="Size of reference genome" optional="True"  argument="--est-ref-size"
            help="Estimated reference genome size (in bp) for computing NGx statistics, if known. This value will be used only if a reference genome file is not specified "/>
        <param name="input_ref" type="data" format="fasta"  label="Reference File" optional="True"  argument="-R"
            help="Many metrics can't be evaluated without a reference. If this is omitted, QUAST will only report the metrics that can be evaluated without a reference."/>
        <param name="annot" type="data" format="txt" label="Gene Annotations" optional="True" argument="-G"
            help="File with gene positions in the reference genome. "/>
        <param name="input_operon" type="data" format="txt" label="Operon Annotations"  optional="True" argument="-O"
            help="File with operon positions in the reference genome"/>
        <param name="gene_selection" type="select" label="Type of organism">
            <option value="prokaryotes">Prokaryotes</option>
            <option value="eukaryote">Eukaryote</option>
            <option value="metagenes">Metagenomes</option>
        </param>
        <param name="min_contig" type="integer" value="500" label="Lower Threshold"  argument="--min-contig"
            help="Set the lower threshold for a contig length. Shorter contigs won't be taken into account (default is 500)"/>
        <param name="threshold_contig" type="text" value="0,1000" label="Thresholds" argument="--contig-thresholds"
            help="Set the thresholds for contig length. Comma-separated list of contig length thresholds.(default is 0,1000)"/>
    </inputs>
    <outputs>
        <data format="txt" name="quast_out" label="Quast report.txt" from_work_dir="Quast_report.txt"/>
        <data format="tsv" name="quast_tsv" label="Quast report.tsv" from_work_dir="Quast_report.tsv"/>
        <data format="tex" name="quast_tex" label="Quast report.tex" from_work_dir="Quast_report.tex"/>
        <data format="html" name="icarus" label="Icarus Contig size viewer" from_work_dir="Icarus.html"/>
        <data format="html" name="report_html" label="Quast report.html" from_work_dir="Quast_report.html"/>
    </outputs>
    <tests>
        <test>
            <param name="input" value="contigs_1.fasta"/>
            <param name="input.type_file" value="contig"/>
            <param name="gene_selection" value="prokaryotes"/>
            <output name="quast_tsv" file="Quast_report.tsv" lines_diff="4"/>
        </test>
    </tests>
    <help>
<![CDATA[
**Description**

Galaxy tool wrapper for the QUAST tool. Quast stands for QUality ASsessment Tool. It evaluates genome assemblies by computing various metrics.

-----

**Inputs and Outputs**

- Input:
    + The tool accepts assemblies and references in FASTA format.
    + The tool accepts annotation and operon files in:
        + GFF, versions 2 and 3 (note: feature/type field should be either "gene" or "operon")
        + the format used by NCBI for genes ("Summary (text)");
        + four tab-separated columns: sequence name, gene/operon id, start position, end position

- Output:
    + An assessment summary in plain text format
    + An assessment summary in tabulation separated values format
    + An assessment summary in LateX format
    + An assessment summary in HTML format
    + An HTML view of contig sizes wit Icarus


    ]]>
    </help>

    <citations>
        <citation type="bibtex">@ARTICLE{a1,
            author = {Alexey Gurevich, Vladislav Saveliev, Nikolay Vyahhi, Glenn Tesler},
            title = {QUAST: quality assessment tool for genomce assemblies, Bioinformatics (2013) 29 (8): 1072-1075}
        }</citation>
        <citation type="bibtex">@misc{quast41,
            title = {{Quast} v4.1},
            howpublished = {http://bioinf.spbau.ru/quast},
            note = {Released May 2016}}
        }</citation>
    </citations>
</tool>