annotate salsa2.xml @ 2:ab5b7f6b7198 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/salsa2 commit 59c5c8edd0c2080d7685e312fee19de1da75097e"

<tool id="salsa" name="SALSA" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="20.01">
    <description>scaffold long read assemblies with Hi-C</description>
    <xrefs>
        <xref type="bio.tools">SALSA</xref>
    </xrefs>
    <macros>
        <token name="@TOOL_VERSION@">2.3</token>
        <token name="@VERSION_SUFFIX@">1</token>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">salsa2</requirement>
        <requirement type="package" version="1.11">samtools</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
ln -s '$fasta_in' input.fasta &&
samtools faidx input.fasta &&

run_pipeline.py
-a '$fasta_in'
-l input.fasta.fai
#if $enzyme_conditional.enzyme_options == 'preconfigured':
    #if $enzyme_conditional.preconfigured_enzymes == 'dovetail'
        -e 'GATC'
    #else if $enzyme_conditional.preconfigured_enzymes == 'arima1'
        -e 'GATC,GANTC'
    #else
        -e 'GATC,GANTC,CTNAG,TTAA'
    #end if
#else:
    -e '${enzyme_conditional.manual_enzyme}'
#end if
-b '$bed_file'
#if str($cutoff):
    -c '$cutoff'
#end if
#if $gfa_file:
    -g '$gfa_file'
#end if
#if $iter:
    -i '$iter'
#end if
-o ./out
    ]]></command>
    <inputs>
        <param name="fasta_in" type="data" format="fasta" label="Initial assembly file" help="Headers must not contain ':'."/>
        <param name="bed_file" type="data" format="bed" label="Bed alignment" help="Sorted by read names"/>
        <param name="cutoff" argument="-c" type="integer" min="1" label="Cutoff" optional="true" help="Minimum contig length to scaffold."/>
        <param name="gfa_file" argument="-g" type="data" format="gfa1,gfa2" optional="true" label="Sequence graphs" help="Sequence graphs encoded in GFA format."/>
        <conditional name="enzyme_conditional">
            <param name="enzyme_options" type="select" label="Enzyme selection" help="TODO">
                <option value="preconfigured">Preconfigured restriction enzymes</option>
                <option value="specific">Enter a specific sequence</option>
            </param>
            <when value="preconfigured">
                <param name="preconfigured_enzymes" type="select" multiple="true" label="Preconfigured enzymes">
                    <option value="dovetail">Dovetail Chicago, Dovetail Hi-C or Phase: GATC</option>
                    <option value="arima1">Arima Hi-C 1.0: GATC, GANTC</option>
                    <option value="arima2">Arima Hi-C 2.0: GATC, GANTC, CTNAG, TTAA</option>
                </param>
            </when>
            <when value="specific">
                <param name="manual_enzyme" argument="-e" type="text" label="Restriction enzyme sequence(s)"
                    help="Restriction enzyme sequence. If multiple were used, include all as a comma separated list without spaces (ex. 'GATC,AAGCTT').">
                    <validator type="expression" message="Only alphabetical letters and the comma can be used in to define restriction enzym sequences.">value.replace(',', '').isalpha()</validator>
                </param>
            </when>
        </conditional>
        <param name="iter" argument="-i" type="integer" min="0" label="Iterations" optional="true" help="Number of iterations to run"/>
    </inputs>
    <outputs>
        <data name="scaffolds_fasta" format="fasta" from_work_dir="out/scaffolds_FINAL.fasta" label="${tool.name} on ${on_string}: FASTA assembly"/>
        <data name="scaffolds_agp" format="tabular" from_work_dir="out/scaffolds_FINAL.agp" label="${tool.name} on ${on_string}: agp output"/>
    </outputs>
    <tests>
        <test>
            <param name="fasta_in" value="test.fasta"/>
            <param name="length" value="test.fai"/>
            <param name="bed_file" value="test.bed"/>
            <param name="gfa_file" value="test.gfa1"/>
            <conditional name="enzyme_conditional">
                <param name="enzyme_options" value="specific"/>
                <param name="manual_enzyme" value="GATC,GANTC"/> 
            </conditional>
            <param name="enzyme" value="GATC,GANTC"/> 
            <param name="cutoff" value="1000"/>
            <param name="iter" value="3"/>
            <output name="scaffolds_fasta" file="out.fasta"/>
            <output name="scaffolds_agp" file="out.agp"/>
        </test>
        <!--Test manual enzyme-->
        <test>
            <param name="fasta_in" value="test.fasta"/>
            <param name="bed_file" value="test.bed"/>
            <param name="gfa_file" value="test.gfa1"/>
            <conditional name="enzyme_conditional">
                <param name="enzyme_options" value="specific"/>
                <param name="manual_enzyme" value="GATC,GANTC"/> 
            </conditional>
            <param name="cutoff" value="1000"/>
            <param name="iter" value="3"/>
            <output name="scaffolds_fasta" file="out.fasta"/>
            <output name="scaffolds_agp" file="out.agp"/>
        </test>
        <!--Test predefined enzyme-->
        <test>
            <param name="fasta_in" value="test.fasta"/>
            <param name="bed_file" value="test.bed"/>
            <param name="gfa_file" value="test.gfa1"/>
            <conditional name="enzyme_conditional">
                <param name="enzyme_options" value="preconfigured"/>
                <param name="preconfigured_enzymes" value="arima1"/> 
            </conditional>
            <param name="cutoff" value="1000"/>
            <param name="iter" value="3"/>
            <output name="scaffolds_fasta" file="out.fasta"/>
            <output name="scaffolds_agp" file="out.agp"/>
        </test>
    </tests>
    <help><![CDATA[
**What is does**

SALSA (Simple AssembLy ScAffolder) is a scaffolding tool based on a computational method that exploits the genomic proximity
information in Hi-C data sets for long range scaffolding of de novo genome assemblies.

    ]]></help>
    <citations>
        <citation type="doi">10.1101/261149</citation>
        <citation type="doi">10.1186/s12864-017-3879-z</citation>
    </citations>
</tool>