view bcftools_norm.xml @ 25:dfba0f0b3edd draft

Uploaded

<tool id="bcftools_norm" name="BCFtools norm" version="1.0.0">
    <description>Left-align and normalize indels, check if REF alleles match reference, change multiallelic format</description>
    <requirements>
        <requirement type="package" version="1.3">bcftools</requirement>
    </requirements>
    <command>
        #if str( $input_file_index ) != "None":
            ln -s -f $input input_file.${input.ext} &amp;&amp;
            ln -s -f $input_file_index input_file.${input.ext}.tbi &amp;&amp;
        #end if
        #if str( $fasta_ref_cond.set_fasta_ref ) == "True":
            #if str( $fasta_ref_cond.fasta_ref_index ) != "None":
                ln -s -f $fasta_ref_cond.fasta_ref fasta_ref_file.${fasta_ref.ext}.fa &amp;&amp;
                ln -s -f $fasta_ref_cond.fasta_ref_index fasta_ref_file.${fasta_ref.ext}.fa.fai &amp;&amp;
            #end if
        #end if
        bcftools norm                    
        --output "${output_file}"
        --output-type "${output_type}"
        #if str( $regions_file ) != "None":
            --regions-file "${regions_file}"
        #end if
        #if str( $regions ) != "":
            --regions "${regions}"
        #end if
        #if str( $targets_file ) != "None":
            --targets-file "${targets_file}"
        #end if
        #if str( $targets ) != "":
            --targets "${targets}"
        #end if
        #if str( $check_ref_cond.set_check_ref ) == "True":
            #if str( $check_ref_cond.warn ) == "true":
                #if str( $check_ref_cond.check_ref ) == "None":
                    --check-ref "w"
                #else
                    --check-ref "w${check_ref_cond.check_ref}"
                #end if
            #else
                --check-ref "${check_ref_cond.check_ref}"
            #end if
        #end if
        #if str( $fasta_ref_cond.set_fasta_ref ) == "True":
            #if str( $fasta_ref_cond.fasta_ref_index ) != "None":
                --fasta-ref "fasta_ref_file.${fasta_ref.ext}.fa"
            #else
                --fasta-ref "${fasta_ref}"
            #end if
        #end if
        #if str( $multiallelics_cond.set_multiallelics ) == "True":
            #if str( $multiallelics_cond.minus_or_plus ) == "minus":
                #if str ( $multiallelics_cond.multiallelics ) != "None":
                    --multiallelics "-${multiallelics_cond.multiallelics}"
                #else
                    --multiallelics "-"
                #end if
            #else
                #if str ( $multiallelics_cond.multiallelics ) != "None":
                    --multiallelics "+${multiallelics_cond.multiallelics}"
                #else
                    --multiallelics "+"
                #end if
            #end if
        #end if
        #if str( $rm_dup ) != "None":
            --rm-dup "${rm_dup}"
        #end if
        ${do_not_normalize}
        ${strict_filter}
        #if str( $site_win ) != "":
            --site-win "${site_win}"
        #end if
        #if str( $input_file_index ) != "None":
            input_file.${input.ext}
        #else
            $input
        #end if
    </command>
    <inputs>
        <param name="input" type="data" format="vcf,bcf,bgzip" label="VCF or BCF input file" help="Input file can optionally be compressed and indexed using Bgzip and Tabix Galaxy tools. In this case, select bgzipped file here and index below."/>
        <param name="input_file_index" type="data" optional="true" label="Input file index" help="Tabix-generated index for input file. Run Tabix Galaxy tool on bgzipped input file to create index."/>
        <param name="output_type" type="select" label="Output data type" help="-O">
            <option value="v" selected="true">VCF</option>
            <option value="b">BCF</option>
        </param>
        <param name="regions_file" type="data" optional="true" label="Regions file" help="-R. Regions specified in a VCF, BED, or tab-delimited file with columns CHROM, POS, and, optionally, POS_TO."/>
        <param name="regions" type="text" size="80" optional="True" label="Regions list" help="-r. Comma-separated list of regions. Format: chr|chr:pos|chr:from-to|chr:from-[,...]">
            <sanitizer invalid_char="">
                <valid initial="string.digits"><add value="CHRchr-:,"/> </valid>
            </sanitizer>
        </param>
        <param name="targets_file" type="data" optional="true" label="Targets file" help="-T. Targets specified in a VCF, BED, or tab-delimited file with columns CHROM, POS, and, optionally, POS_TO."/>
        <param name="targets" type="text" size="80" optional="True" label="Targets list" help="-t. Comma-separated list of targets. Format: [^]chr|chr:pos|chr:from-to|chr:from-[,...]">
            <sanitizer invalid_char="">
                <valid initial="string.digits"><add value="CHRchr-:,^"/> </valid>
            </sanitizer>
        </param>
        <conditional name="fasta_ref_cond">
            <param name="set_fasta_ref" type="boolean" checked="False" label="Specify FASTA reference sequence" help="-f" />
            <when value="true">
                <param name="fasta_ref" type="data" label="FASTA reference file" help="Reference sequence. Using this option turns on left-alignment and normalization. Set corresponding index below." />
                <param name="fasta_ref_index" type="data" optional="True" label="FASTA reference index" help="Run Faidx (SAMtools) Galaxy tool on above reference file to create index. Errors can arise if this index is not specified." />
            </when>
        </conditional>
        <param name="do_not_normalize" type="boolean" checked="False" truevalue="--do-not-normalize" falsevalue="" label="Turn off indel normalization" help="-N. Indel normalization automatically occurs when FASTA reference is used. Use this option to disable this behavior." />
        <conditional name="check_ref_cond">
            <param name="set_check_ref" type="boolean" checked="False" label="Check REF" help="-c. Choose action when an incorrect or missing REF allele is encountered." />
            <when value="true">
                <param name="check_ref" type="select" optional="True" label="Action" help="Action when an incorrect or missing REF allele is encountered.">
                    <option value="e">Exit program</option>
                    <option value="x">Exclude incorrect/missing sites</option>
                    <option value="s">Set or fix incorrect/missing sites (requires FASTA reference)</option>
                </param>
                <param name="warn" type="boolean" checked="False" label="Warn" help="Warn when an incorrect or missing REF allele is encountered." />
            </when>
        </conditional>
        <param name="rm_dup" type="select" optional="True" label="Remove duplicates" help="-d. For given variant type, if a record is present in multiple files, output only the first instance.">
            <option value="snps">SNPs</option>
            <option value="indels">Indels</option>
            <option value="both">Both (SNPs and indels)</option>
            <option value="any">Any</option>
        </param>
        <conditional name="multiallelics_cond">
            <param name="set_multiallelics" type="boolean" checked="False" label="Modify multiallelic sites" help="-m. Split multiallelic sites into biallelic records, or join biallelic sites into multiallelic records." />
            <when value="true">
                <param name="minus_or_plus" type="select" label="Split or join">
                    <option value="minus" selected="true">Split multiallelic sites into biallelic records</option>    
                    <option value="plus">Join biallelic sites into multiallelic records</option>
                </param>      
                <param name="multiallelics" type="select" optional="True" label="Sites to split/join" help="">
                    <option value="snps">SNPs (only SNP records are split/joined)</option>
                    <option value="indels">Indels (only indel records are split/joined)</option>
                    <option value="both">Both (SNPs and indels are merged separately into two records)</option>
                    <option value="any">Any (SNPs and indels are merged into a single record)</option>
                </param>
            </when>
        </conditional>
        <param name="strict_filter" type="boolean" checked="False" truevalue="--strict-filter" falsevalue="" label="Strict PASS filter" help="-s. When merging biallelics into multiallelics, merged site has value PASS only if all sites being merged have value PASS." />
        <param name="site_win" type="integer" optional="true" label="Position window" help="-w. Maximum distance between two records to consider when locally sorting variants which changed position during the realignment." />
    </inputs>
    <outputs>
        <data name="output_file" format="vcf" label="${tool.name} on ${on_string}">
            <change_format>
                <when input="output_type" value="b" format="bcf" />
            </change_format>
        </data>
    </outputs>
    <help>
**About this tool**

**BCFtools norm**: Left-align and normalize indels, check if REF alleles match the reference, split multiallelic sites into multiple rows, recover multiallelics from multiple rows. Left-alignment and normalization will only be applied if the --fasta-ref option is supplied.

Please see https://samtools.github.io/bcftools/bcftools.html for more info on options.
  </help>
</tool>