<tool id="dr_disco_integrate" name="Dr. Disco (integrate)" version="@TOOL_VERSION@">
    <description>Merges corresponding genomic breaks and exon-to-exon junctions</description>
    <macros>
        <import>macros.xml</import>
    </macros>

    <expand macro="requirements"/>
    <expand macro="version_command"/>
    
    <command detect_errors="exit_code"><![CDATA[
        dr-disco
            integrate
                #if $gtf:
                    --gtf '${gtf}'
                #end if

                '${dr_disco_classify_output}'
                '${dr_disco_integrate_output}'
    ]]></command>

    <inputs>
        <param name="dr_disco_classify_output" type="data" format="tabular" label="Discordant alignment file of STAR (processed with dr-disco fix)" />
        <param argument="--gtf" type="data" format="gtf" optional="True" label="GTF file (Optional: for predicting frame shifts)" help="This GTF file requires the following attributes: gene_name, transcript_id and transcript_version"/>
    </inputs>
    
    <outputs>
        <data name="dr_disco_integrate_output" format="tabular" label="${tool.name} on ${dr_disco_classify_output.name}" />
    </outputs>
    
    <tests>
        <test>
            <param name="dr_disco_classify_output" value="integrate_test_terg_s041.in.dbed" />
            <param name="gtf" value="integrate_tmprss-erg.hg38.gtf" />

            <output name="dr_disco_integrate_output" file="integrate_test_terg_s041.out.txt" />
        </test>
        <test>
            <param name="dr_disco_classify_output" value="integrate_test_terg_s041.in.dbed" />
            <!--
            <param name="gtf" value="integrate_tmprss-erg.hg38.gtf" />
            -->

            <output name="dr_disco_integrate_output" file="integrate_test_terg_s041.out.no-gtf.txt" />
        </test>
    </tests>
    
    <help><![CDATA[

**What it does**

* Step 4 in dr-disco pipeline

In this step the predicted junctions are integrated with a gene annotation. The junctions are provided names like 'TMPRSS2->ERG' and frameshift are provided. Also events that belong to each other, thus the intronic and exonic junctions that belong to the same genomic event, are given the same identifier.

**Attribution**

This work is part of Youri Hoogstrate's PhD thesis.

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