annotate annotate_variants.xml @ 2:9c5c2d4cf379 draft

planemo upload for repository https://github.com/wm75/mimodd_galaxy_wrappers commit 528bcf3b769c7c73f119b2a176d19071f9ef5312

<tool id="mimodd_annotate" name="MiModD Variant Annotation" 
version="@MIMODD_WRAPPER_VERSION@">
    <description>
    with functional effects on genes and transcripts
    </description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements" />
    <expand macro="stdio" />
    <expand macro="version_command" />
    <command><![CDATA[
  	mimodd annotate
	  '$inputfile' '$genome.annotation_source'
	  #if $snpeff_settings.stats:
	    --stats '$summary_file'
	  #end if
	  #if $codon_tables.genome_codon_table.strip() or len($codon_tables.table_specs) > 0:
	    --codon-tables
	    #if $codon_tables.genome_codon_table.strip():
	      '$codon_tables.genome_codon_table'
	    #end if
	    #for $table_spec in $codon_tables.table_specs:
	      '$table_spec.contig:$table_spec.table'
	    #end for
	  #end if
	  #if $snpeff_settings.ud:
	    --ud ${snpeff_settings.ud}
	  #end if
	  #if $snpeff_settings.filter:
	    #echo ' '.join(str($snpeff_settings.filter).split(','))
	  #end if
	  --ofile '$ofile'
	  --verbose --quiet
    ]]></command>

    <inputs>
        <param name="inputfile" type="data" format="vcf"
        label="VCF input file to be annotated" />
        <conditional name="genome">
            <param name="selection_method" type="select"
            label="Select SnpEff Genome Database">
                <option value="select_from_list">from a list of installed databases</option>
                <option value="select_by_name">by name</option>
            </param>
            <when value="select_from_list">
                <param name="genome_list" type="data" format="tabular" optional="true"
                label="List of available SnpEff Genomes"
                help="Select a dataset generated with the MiModD List Installed SnpEff Genomes tool to be able to choose from the genome databases listed in it." />
                <param name="annotation_source" type="select"
                label="SnpEff Genome to use"
                help="Select a genome annotation database for your organism from the ones in the list.">
                    <options from_dataset="genome_list">
                        <column name="name" index="0"/>
                        <column name="value" index="1"/>
                        <filter type="unique_value" name="unique_names" column="0"/>
                    </options>
                </param>
            </when>
            <when value="select_by_name">
                <param name="annotation_source" type="text"
                label="Name of an installed SnpEff genome"
                help="Your input needs to exactly match the name of an installed SnpEff database.">
                    <validator type="expression" message="A SnpEff genome is required to run this tool.">value.strip()</validator>
                    <expand macro="lex_sam_header" message="Non-ASCII characters are not valid in SnpEff genome names." />
                </param>
            </when>
        </conditional>
        <section name="codon_tables" title="Codon Table Configuration" expanded="false">
            <param name="genome_codon_table" type="text"
            label="Genome codon table"
            help="If this genome uses a non-standard codon table and this fact is not declared in the SnpEff config file, specify the table here. The name you provide must match exactly the name of a codon table known to SnpEff.">
                <expand macro="lex_sam_header" message="Non-ASCII characters are not valid in codon table names." />
            </param>
            <repeat name="table_specs" default="0" min="0"
            title="Codon tables to use for specific contigs"
            help="Use if certain contigs in the reference genome use a distinct codon table. A typical example is a mitochondrial genome contained as a contig in the reference genome. Contig/codon table assignments declared here will take precedence over a genome codon table set above.">
                <param name="contig" type="text" label="contig or chromosome name">
                    <validator type="expression" message="This is a required field for a contig/codon table assignment.">value.strip()</validator>
                    <expand macro="lex_sam_header" message="Non-ASCII characters are not allowed in contig/chromosome names" />
                </param>
                <param name="table" type="text" label="uses codon table">
                    <validator type="expression" message="This is a required field for a contig/codon table assignment.">value.strip()</validator>
                    <expand macro="lex_sam_header" message="Non-ASCII characters are not valid in codon table names." />
                </param>
            </repeat>
        </section>
    	<section name="snpeff_settings" title="More SnpEff options" expanded="false">
            <param name="stats" type="boolean" checked="false"
            label="Have SnpEff produce a summary file of results" />
            <param name="ud" type="integer" optional="true" 
            label="upstream downstream interval length (default = 5000 bases)"
            help="specify the upstream/downstream interval length. Variants more than INTERVAL nts from the next annotated gene will be considered intergenic"/>
            <param name="filter" type="select" display="checkboxes" multiple="true"
            label="Disable select types of annotations"
            help="The corresponding variants will still be retained (without annotations) in the output">
                <option value="--no-downstream">Do not annotate DOWNSTREAM changes</option>
                <option value="--no-upstream">Do not annotate UPSTREAM changes</option>
                <option value="--no-intergenic">Do not annotate INTERGENIC changes</option>
                <option value="--no-intron">Do not annotate INTRON changes</option>
                <option value="--no-utr">Do not annotate 5_PRIME_UTR or 3_PRIME_UTR changes</option>
            </param>
        </section>
    </inputs>

    <outputs>
        <data name="ofile" format="vcf"
        label="Variants annotated with MiModd from ${on_string}" />
        <data name="summary_file" format="html"
        label="Summary of MiModD variant annotation of ${on_string}">
            <filter>snpeff_settings['stats']</filter>
        </data>
    </outputs>

    <help><![CDATA[
.. class:: warningmark

This tool requires the variant annotation software SnpEff, which is not part of
the MiModD package! It also needs the correct SnpEff genome database for your
organism and reference sequence version installed with SnpEff.

.. class:: infomark

   **What it does**

This tool uses SnpEff to annotate the variants in the VCF input dataset with
the effects that these variants have on known genomic features, i.e. on genes
and transcripts.

To do its job, SnpEff requires a database of known genome annotations for your
organism of interest. If you know the exact name of the genome database that
you want to use, you can enter it directly. Otherwise, you should use the
MiModD *List Installed SnpEff Genomes* tool first to generate a new dataset in
your history that lists all SnpEff genome databases currently available to
SnpEff on the system, then when you select this dataset as the *List of
available SnpEff Genomes* above, you will be able to choose the SnpEff Genome
from a dropdown list.

The tool will create a new dataset in VCF format with the annotations added to
the corresponding variant records. Do not expect this format to be very
readable, but use the *MiModD Variant Reports* tool to turn it into a human
friendly form.

In addition, you can instruct the tool to produce an additional dataset with a
summary of the variants and their effects in html format.

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