annotate repeat_annotate_custom.xml @ 12:755a4d643184 draft default tip

planemo upload commit a61591d548f42ff417781e7fe7418dc2901ccc23

<tool id="repeat_annotate" name="Library Based Assembly Annotation" version="0.1.7" python_template_version="3.5">
    <requirements>
        <requirement type="package" version="4.1.2.p1">repeatmasker</requirement>
        <requirement type="package">bioconductor-rtracklayer</requirement>
    </requirements>
    <required_files>
        <include type="literal" path="clean_rm_output.R"/>
    </required_files>
    <command detect_errors="exit_code"><![CDATA[

        #if $sensitivity.value == 'default':
        RepeatMasker -dir \$(pwd) '$input' -pa 32 -lib '$repeat_library' -xsmall ${nolow} -no_is -e ncbi
        #else:
        RepeatMasker -dir \$(pwd) '$input' -pa 32 -lib '$repeat_library' -xsmall ${nolow} -no_is -e ncbi $sensitivity
        #end if
        &&
        ls -l * >&2 &&
        cp `basename $input`.out  $output2
        &&
        Rscript ${__tool_directory__}/clean_rm_output.R $output2 $output1

        ]]></command>
    <inputs>
        <param type="data" name="input" format="fasta" label="Genome assembly to annotate" />
        <param type="data" name="repeat_library" format="fasta" label="RepeatExplorer based Library of Repetitive Sequences"
               help="custom database of repetitive sequences should be provided in fasta format. Sequence header should specify repeat class:
                     >sequence_id#classification_level1/classification_level2/..." />
        <param type="select" label="sensitivity" name="sensitivity" >
            <option value="default" selected="true" >Default sensitivity </option>
            <option value="-s">Slow search, more sensitive </option>
            <option value="-q">Quick search, less sensitive </option>
        </param>
        <param argument="-nolow" type="boolean" truevalue="" falsevalue="-nolow" checked="true" label="Perform masking of low complexity and simple repeats" />
    </inputs>
    <outputs>
        <data name="output1" format="gff3"  label="Repeat Annotation (GFF3) on ${on_string}"/>
        <data name="output2" format="tabular" label="Raw output from RepeatMasker on ${on_string}" />
    </outputs>
    <help><![CDATA[
        This tools uses RepeatMasker to annotate repetitive sequences in the genome assemblies using custom library of repeats created either from RepeatExplorer output or from DANTE_LTR transposable element library.
        Library of repeats can be created from RepeatExplorer output from contigs and TAREAN consensus sequences.
        Fasta formatted library of repeats must contain header containing  information about classification of repeats as **>sequence_id#classification_level1/classification_level2/...**

        Classification in RepeatExplorer based library follows predetermined classification levels. User can however specify additional classification levels or custom classifications.
        Conflicts in annotations are resolved based on classification hierarchy.
    ]]></help>
</tool>