view vcftools_compare.xml @ 0:937918765b6e draft default tip

Uploaded

<tool id="vcftools_compare" name="Compare" version="0.1">
    <description>multiple VCF datasets</description>

    <requirements>
        <requirement type="package">tabix</requirement>
        <requirement type="package" version="0.1.11">vcftools</requirement>
    </requirements>
    
    <command>
        ## Preprocessing for each dataset.
        #set dataset_names = []
        #for $input in $inputs:
            ## Sort file.                                     
            sort -k1,1 -k2,2n ${input.file} > ${input.name}.vcf.sorted ;

            ## Compress.
            bgzip ${input.name}.vcf.sorted ;

            ## Index.
            tabix -p vcf ${input.name}.vcf.sorted.gz ;

            #silent dataset_names.append( str($input.name) + '.vcf.sorted.gz' )
        #end for

        ## Compare and use sed to simplify output.
        vcf-compare -w ${window} ${ignore_indels} #echo ' '.join( dataset_names )# | sed 's/\.vcf\.sorted\.gz//g' > ${output}
    </command>
    <inputs>
        <repeat name="inputs" title="Datasets to Compare" min="2">
            <param name="name" label="Dataset name" type="text" />
            <param name="file" label="Dataset" type="data" format="vcf"/>
        </repeat>
        <param name="window" label="Comparison window" type="integer" min="0" value="0" help="In repetitive sequences, the same indel can be called at different positions. Consider records this far apart as matching (be it a SNP or an indel)."/>

        <param name="ignore_indels" label="Ignore indels" type="select" help="Exclude sites containing indels from genotype comparison">
            <option value="" selected="True">No</option>
            <option value="--ignore-indels">Yes</option>
        </param>
    </inputs>

    <outputs>
        <data name="output" format="tabular"/>
    </outputs>

    <stdio>
        <regex match=".*" source="both" level="log" description="tool progress"/>
    </stdio>

    <tests>
        <!-- Cannot specify multiple repeats in test framework right now, so no tests possible. -->
    </tests>

    <help>
        Please see the VCFtools `documentation`__ for help and further information.

        .. __: http://vcftools.sourceforge.net/docs.html
    </help>
</tool>