annotate mosdepth.xml @ 2:efd53f6ceb16 draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/mosdepth commit 37a76b08262eee35b2bdb2604d6a7beda046fe68

<tool id="mosdepth" name="mosdepth" version="@TOOL_VERSION@+galaxy0"  profile="21.05">
    <description>- fast and flexible depth coverage calculation</description>
    <macros>
        <xml name="bam_identifier">
            <!-- this is based on section 1.2.1 of the SAM/BAM specification https://samtools.github.io/hts-specs/SAMv1.pdf -->
            <sanitizer invalid_char="'">
                <valid initial="string.ascii_letters,string.digits,string.punctuation">
                    <remove value="{" />
                    <remove value="}" />
                    <remove value="=" />
                    <remove value="*" />
                    <remove value="`" />
                    <remove value=")" />
                    <remove value="&lt;" />
                    <remove value="'" />
                    <remove value='"' />
                    <remove value="(" />
                    <remove value="\" />
                    <remove value="[" />
                    <remove value="&gt;" />
                    <remove value="]" />
                    <remove value="," />
                    <yield />
                </valid>
            </sanitizer>
            <validator type="regex">([0-9A-Za-z!#$%&amp;+./:;?@^_|~-][0-9A-Za-z!#$%&amp;*+./:;=?@^_|~-]*)?</validator>
        </xml>
        <token name="@TOOL_VERSION@">0.3.5</token>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">mosdepth</requirement>
        <requirement type="package" version="1.13">gzip</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        #if $advanced.options == "yes" and len($advanced.quantize)
            #set $quantize_depths = []
            #for $i, $quant_group in enumerate($advanced.quantize)
                #set $quantize_depths = $quantize_depths + [str ($quant_group.quant_group_mindepth)]
                #set $quant_group_name = str($quant_group.quant_group_name)
                export MOSDEPTH_Q${i}=$quant_group.quant_group_name &&
            #end for
            #set $quant_groups_depths_str = ":".join($quantize_depths) + ':'
        #end if
        ln -s '$input_alignment' input_alignment &&
        #if $input_alignment.is_of_type('bam'):
            ln -s '${input_alignment.metadata.bam_index}' input_alignment.bai &&
        #elif $input_alignment.is_of_type('cram'):
            ln -s '${input_alignment.metadata.cram_index}' input_alignment.crai &&
        #end if
        mosdepth -t \${GALAXY_SLOTS:-1} 
            #if $window.window_mode == 'window'
                --by $window.window_size
            #elif $window.window_mode == 'bed'
                --by '$window.region_file'
            #end if
            $per_base_coverage 
            #if $advanced.options == "yes"
                #if str($advanced.chrom)
                --chrom '$advanced.chrom'
                #end if
                #if $advanced.exclude_flag
                    --flag $advanced.exclude_flag
                #end if
                #if $advanced.include_flag
                    --include-flag $advanced.include_flag
                #end if
                --mapq $advanced.mapq
                $advanced.no_fast
                #if str($advanced.thresholds)
                    --thresholds '$advanced.thresholds'
                #end if
                $advanced.use_median
                #if str($advanced.read_groups)
                    --read-groups '$read_groups'
                #end if
                #if len($advanced.quantize)
                --quantize '$quant_groups_depths_str'
                #end if
            #end if 
            output input_alignment
            #if $per_base_coverage
                && gunzip output.per-base.bed.gz
            #end if
            #if $window.window_mode == "bed" or $window.window_mode == "window"
                && gunzip output.regions.bed.gz
            #end if
            #if $advanced.options == "yes"
                #if str($advanced.thresholds)
                    && gunzip output.thresholds.bed.gz
                #end if
                #if len($advanced.quantize)
                    && gunzip output.quantized.bed.gz
                #end if
            #end if
    ]]></command>
    <inputs>
        <param name="input_alignment" type="data" format="bam,cram" label="Input BAM (or CRAM)" help="BAM or CRAM format mapped reads" />
        <param argument="--no-per-base" name="per_base_coverage" type="boolean" truevalue="" falsevalue="--no-per-base" checked="false" label="Output per base depth" help="Skipping depth output for individual bases substantially decreases runtime" />
        <conditional name="window">
            <param name="window_mode" type="select" label="Compute depth by region" help="A BED file or window size can be provided to compute average depth by region">
                <option value="no">Don't compute depth by region</option>
                <option value="window">Compute average depth using a fixed size window</option>
                <option value="bed">Compute depth in regions specified by a BED file</option>
            </param>
            <when value="no">                
            </when>
            <when value="window">
                <param name="window_size" type="integer" min="2" value="400" label="Window size for region calculation" help="Average depth will be computed for non-overlapping regions of this size" />
            </when>
            <when value="bed">
                <param name="region_file" type="data" format="bed" label="BED file specifying regions" help="Average depth will be computed for regions specified in the BED file. If the BED file has a 4th column, it is used for the region name" />
            </when>
        </conditional>
        <conditional name="advanced">
            <param name="options" type="select" label="Advanced Options">
                <option value="no" selected="true">Hide Advanced Options</option>
                <option value="yes">Show Advanced Options</option>
            </param>
            <when value="no">
            </when>
            <when value="yes">
                <param argument="--chrom" type="text" label="Restrict depth calculations to chromosome" optional="true" help="Enter a chromosome name to only count depth for reads mapping to this chromosome">
                    <expand macro="bam_identifier" />
                </param>
                <param argument="--flag" name="exclude_flag" type="integer" min="0" value="0" label="Exclude reads with these bits set in FLAG" help="Exclude reads where any of these bits are set in the value of the read flags column" />
                <param argument="--include-flag" type="integer" min="1" optional="true" label="Only include reads with these bits set in FLAG" help="Only include reads where any of these bits are set in the value of the read flags column" />
                <param argument="--fast-mode" name="no_fast" type="boolean" truevalue="" falsevalue="-x" label="Disable fast mode" help="Fast mode doesn't look up internal cigar operations or correct mate overlaps. Disabling it is not recommended" />
                <param argument="--mapq" type="integer" min="0" value="0" label="Minimum mapping quality" help="Reads with a mapping quality lower than this value are ignored" />
                <param argument="--thresholds" type="text" value="" label="Specify thresholds for output when using region output" help="The number of bases covered by these thresholds will be reported. Multiple thresholds can be separated by commas">
                    <sanitizer invalid_char="">
                            <valid initial="string.digits">
                            <add value="," />
                        </valid>
                    </sanitizer>
                    <validator type="regex">[0-9,]*</validator>
                </param>
                <param argument="--use-median" type="boolean" truevalue="--use-median" falsevalue="" checked="false" label="Use median depth of region" help="When computing depth of region, use median of depth rather than mean" />
                <param argument="--read-groups" type="text" value="" label="" help="only calculate depth for these comma-separated read groups IDs">
                    <!-- while read group IDs are not well specified in the BAM specification, they appear to be like IDs but with spaces allowed -->
                    <expand macro="bam_identifier">
                        <add value=" " />
                    </expand>
                </param>
                <repeat name="quantize" title="Read depth thresholds for depth-defined regions">
                    <param name="quant_group_mindepth" type="integer" min="0" value="0" label="Minimum depth for depth-defined region" help="Regions with this depth or lower will be included in this depth-defined region" />
                    <param name="quant_group_name" type="text" label="Depth-defined region name" help="Name include in BED output for regions in this depth-defined region">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters,string.digits">
                                <add value="_" />
                            </valid>
                        </sanitizer>
                        <validator type="regex">[0-9a-zA-Z_]*</validator>
                    </param>
                </repeat>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_global_dist" format="tabular" from_work_dir="output.mosdepth.global.dist.txt" label="mosdepth depth distribution on ${on_string}" />
        <data name="output_summary" format="tabular" from_work_dir="output.mosdepth.summary.txt" label="mosdepth summary on ${on_string}">
            <actions>
                <action type="metadata" name="column_names" default="chrom,length,bases,mean,min,max" />
            </actions>
        </data>
        <data name="output_region_dist" format="tabular" from_work_dir="output.mosdepth.region.dist.txt" label="mosdepth region depth distribution on ${on_string}">
            <filter>window['window_mode'] == "bed" or window['window_mode'] == "window"</filter>
        </data>
        <data name="output_per_base_depth" format="bedgraph" from_work_dir="output.per-base.bed" label="mosdepth per base depth on ${on_string}">
            <filter>per_base_coverage</filter>
        </data>
        <data name="output_regions_bed" format="bed" from_work_dir="output.regions.bed" label="mosdepth regions BED on ${on_string}">
            <filter>window['window_mode'] == "bed" or window['window_mode'] == "window"</filter>
        </data>
        <data name="output_quantized_bed" format="bed" from_work_dir="output.quantized.bed" label="mosdepth quantized BED on ${on_string}">
            <filter>advanced['options'] == "yes" and len(advanced['quantize'])</filter>
        </data>
        <data name="output_thresholds_bed" format="bed" from_work_dir="output.thresholds.bed" label="mosdepth thresholds BED on ${on_string}">
            <filter>advanced['options'] == "yes" and advanced['thresholds']</filter>
        </data>
    </outputs>
    <tests>
        <!-- test 1 -->
        <test expect_num_outputs="2">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <output name="output_summary" file="output_summary.tabular" ftype="tabular" />
            <output name="output_global_dist" file="output_distribution.tabular" ftype="tabular" />
        </test>
        <!-- test 2 -->
        <test expect_num_outputs="3">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <param name="per_base_coverage" value="True" />
            <output name="output_per_base_depth" file="output_per_base_depth.bed" ftype="bedgraph" />
        </test>
        <!-- test 3 -->
        <test expect_num_outputs="4">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <conditional name="window">
                <param name="window_mode" value="bed" />
                <param name="region_file" value="input_regions.bed" ftype="bed" />
            </conditional>
            <output name="output_region_dist" file="output_region_distribution_bed.tabular" ftype="tabular" />
            <output name="output_summary" file="output_region_summary.tabular" ftype="tabular" />
            <output name="output_regions_bed" file="output_region_depths.bed" ftype="bed" />
        </test>
        <!-- test 4 -->
        <test expect_num_outputs="4     ">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <conditional name="window">
                <param name="window_mode" value="window" />
                <param name="window_size" value="200" />
            </conditional>
            <output name="output_region_dist" file="output_region_distribution.tabular" ftype="tabular" />
            <output name="output_summary" file="output_region_summary_windowed.tabular" ftype="tabular" />
            <output name="output_regions_bed" file="output_window_depths.bed" ftype="bed" />
        </test>
        <!-- test 5 -->
        <test expect_num_outputs="4">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <conditional name="window">
                <param name="window_mode" value="window" />
                <param name="window_size" value="200" />
            </conditional>
            <conditional name="advanced">
                <param name="options" value="yes" />
                <param name="use_median" value="yes" />
            </conditional>
            <output name="output_region_dist" file="output_region_distribution_median.tabular" ftype="tabular" />
            <output name="output_summary" file="output_region_summary_windowed_median.tabular" ftype="tabular" />
            <output name="output_regions_bed" file="output_window_depths_median.bed" ftype="bed" />
        </test>
        <!-- test 6 -->
        <test expect_num_outputs="3">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <conditional name="advanced">
                <param name="options" value="yes" />
                <repeat name="quantize">
                    <param name="quant_group_mindepth" value="0" />
                    <param name="quant_group_name" value="NO_COVERAGE" />
                </repeat>
                <repeat name="quantize">
                    <param name="quant_group_mindepth" value="1" />
                    <param name="quant_group_name" value="LOW_COVERAGE" />
                </repeat>
                <repeat name="quantize">
                    <param name="quant_group_mindepth" value="20" />
                    <param name="quant_group_name" value="SUFFICIENT_COVERAGE" />
                </repeat>
            </conditional>
            <output name="output_quantized_bed" file="output_quantized.bed" ftype="bed" />
        </test>
        <!-- test 7 -->
        <test expect_num_outputs="5">
            <param name="input_alignment" value="input.bam" ftype="bam" />
            <conditional name="window">
                <param name="window_mode" value="bed" />
                <param name="region_file" value="input_regions.bed" ftype="bed" />
            </conditional>
            <conditional name="advanced">
                <param name="options" value="yes" />
                <param name="thresholds" value="20,40" />
            </conditional>
            <output name="output_thresholds_bed" file="output_thresholds.bed" ftype="bed" />
        </test>
        <!-- test 7 -->
        <test expect_num_outputs="2">
            <param name="input_alignment" value="input.cram" ftype="cram" />
            <output name="output_summary" file="output_summary.tabular" ftype="tabular" />
            <output name="output_global_dist" file="output_distribution.tabular" ftype="tabular" />
        </test>
    </tests>
    <help><![CDATA[
mosdepth_ is a tool for fast and flexible calculation of read depths from BAM or CRAM files.

It can compute:

* mean (or median) depth in fixed sized windows
* mean (or median) depth in regions specified by a BED file
* per base depths
* a histogram of read depths
* the mean or median coverage histogram for windows / regions
* a distribution of proportion of based covered over a particular threshold
* a BED format report on regions that are defined by coverage thresholds

By default only a summary and depth histogram is computed, but the other options mentioned above can
be enabled using different options in the "Compute depth by region" selector and some of the Advanced
options.

.. _mosdepth: https://github.com/brentp/mosdepth
    ]]></help>
    <citations>
        <citation type="doi">10.1093/bioinformatics/btx699</citation>
    </citations>
</tool>