<tool id="dr_disco_classify" name="Dr. Disco (classify)" version="@TOOL_VERSION@">
    <description>Classifies detected break-points in RNA-seq based on corresponding statistics and blacklists</description>

    <macros>
        <import>macros.xml</import>
    </macros>
    
    <expand macro="requirements"/>
    <expand macro="version_command"/>
    
    <command detect_errors="exit_code"><![CDATA[
        dr-disco
            classify
            
                ${only_valid}
                
                #if $blacklist_regions:
                    --blacklist-regions '${blacklist_regions}'
                #end if

                #if $blacklist_junctions:
                    --blacklist-junctions '${blacklist_junctions}'
                #end if
                
                --min-chim-overhang '${min_chim_overhang}'
                
                '${dr_disco_detect_output}'
                '${dr_disco_classify_output}'
    ]]></command>

    <inputs>
        <param name="dr_disco_detect_output" type="data" format="tabular" label="Output of Dr. Disco Detect" />
        <param name="only_valid" argument="--only-valid" truevalue="--only-valid" falsevalue="" type="boolean" />
        
        <param name="blacklist_regions"   argument="--blacklist-regions"   type="data" format="bed"     optional="True" label="Blacklist Regions"   help="List of regions known to be false positives (https://github.com/yhoogstrate/dr-disco/tree/master/share/)" />
        <param name="blacklist_junctions" argument="--blacklist-junctions" type="data" format="tabular" optional="True" label="Blacklist Junctions" help="List of junctions know to be false positives (https://github.com/yhoogstrate/dr-disco/tree/master/share/)" />

        <param name="min_chim_overhang" type="integer" min="-1" value="50" label="Minimum alignment length on each side of the junction." help="May need to be set to smaller values for read lengths smaller than 75bp." argument="--min-chim-overhang" />
    </inputs>
    
    <outputs>
        <data name="dr_disco_classify_output" format="tabular" label="${tool.name} on ${dr_disco_detect_output.name}" />
    </outputs>
    
    <tests>
        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" value="" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-01.out.dbed" ftype="tabular" />
        </test>

        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" value="--only-valid" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-02.out.dbed" />
        </test>

        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" />
            <param name="blacklist_regions" value="blacklist-regions.hg38.bed" ftype="bed" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-03.out.dbed" ftype="tabular" />
        </test>

        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" />
            <param name="blacklist_regions" value="blacklist-regions.hg38.bed" ftype="bed" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-04.out.dbed" />
        </test>

        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" />
            <param name="blacklist_junctions" value="blacklist-junctions.hg38.txt" ftype="tabular" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-05.out.dbed" ftype="tabular" />
        </test>
        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" />
            <param name="blacklist_junctions" value="blacklist-junctions.hg38.txt" ftype="tabular" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-06.out.dbed" />
        </test>
        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" />
            <param name="blacklist_regions" value="blacklist-regions.hg38.bed" ftype="bed" />
            <param name="blacklist_junctions" value="blacklist-junctions.hg38.txt" ftype="tabular" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-07.out.dbed" ftype="tabular" />
        </test>
        <test>
            <param name="dr_disco_detect_output" value="classify_test_16.in.dbed" ftype="tabular" />
            <param name="only_valid" />
            <param name="blacklist_regions" value="blacklist-regions.hg38.bed" ftype="bed" />
            <param name="blacklist_junctions" value="blacklist-junctions.hg38.txt" ftype="tabular" />

            <output name="dr_disco_classify_output" file="classify_test_16.test-08.out.dbed" />
        </test>

    </tests>
    
    <help><![CDATA[

**What it does**

* Step 3 in dr-disco pipeline

This tools filters the detected junctions which are likely to be true or false based on a variety of parameters provided in the input table. This filtering is in particular based on variables derived from the shape of the alignment and the ratio of split and spanning reads.

**Attribution**

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

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