Repository 'samtools_coverage'
hg clone https://toolshed.g2.bx.psu.edu/repos/iuc/samtools_coverage

Changeset 0:90f2dd176d80 (2021-09-28)
Next changeset 1:ae36fab06bc2 (2022-08-15)
Commit message:
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/samtools/sam_to_bam commit 4d0362c336f67d05086298ab4980c1d922d71c6e"
added:
macros.xml
samtools_coverage.xml
test-data/results_1.tabular
test-data/results_2.txt
test-data/results_3.txt
test-data/test_1.bam
test-data/test_2.bam
b
diff -r 000000000000 -r 90f2dd176d80 macros.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/macros.xml Tue Sep 28 15:10:16 2021 +0000
[
b'@@ -0,0 +1,223 @@\n+<macros>\n+    <xml name="requirements">\n+        <requirements>\n+            <requirement type="package" version="@TOOL_VERSION@">samtools</requirement>\n+            <yield/>\n+        </requirements>\n+    </xml>\n+    <token name="@TOOL_VERSION@">1.13</token>\n+    <token name="@PROFILE@">20.05</token>\n+    <token name="@FLAGS@"><![CDATA[\n+        #set $flags = 0\n+        #if $filter\n+            #set $flags = sum(map(int, str($filter).split(\',\')))\n+        #end if\n+    ]]></token>\n+    <token name="@PREPARE_IDX@"><![CDATA[\n+        ##prepare input and indices\n+        ln -s \'$input\' infile &&\n+        #if $input.is_of_type(\'bam\'):\n+            #if str( $input.metadata.bam_index ) != "None":\n+                ln -s \'${input.metadata.bam_index}\' infile.bai &&\n+            #else:\n+                samtools index infile infile.bai &&\n+            #end if\n+        #elif $input.is_of_type(\'cram\'):\n+            #if str( $input.metadata.cram_index ) != "None":\n+                ln -s \'${input.metadata.cram_index}\' infile.crai &&\n+            #else:\n+                samtools index infile infile.crai &&\n+            #end if\n+        #end if\n+    ]]></token>\n+    <token name="@PREPARE_IDX_MULTIPLE@"><![CDATA[\n+        ##prepare input and indices\n+        #for $i, $bam in enumerate( $input_bams ):\n+            ln -s \'$bam\' \'${i}\' &&\n+            #if $bam.is_of_type(\'bam\'):\n+                #if str( $bam.metadata.bam_index ) != "None":\n+                    ln -s \'${bam.metadata.bam_index}\' \'${i}.bai\' &&\n+                #else:\n+                    samtools index \'${i}\' \'${i}.bai\' &&\n+                #end if\n+            #elif $bam.is_of_type(\'cram\'):\n+                #if str( $bam.metadata.cram_index ) != "None":\n+                    ln -s \'${bam.metadata.cram_index}\' \'${i}.crai\' &&\n+                #else:\n+                    samtools index \'${i}\' \'${i}.crai\' &&\n+                #end if\n+            #end if\n+        #end for\n+    ]]></token>\n+    <token name="@PREPARE_FASTA_IDX@"><![CDATA[\n+        ##checks for reference data ($addref_cond.addref_select=="history" or =="cached")\n+        ##and sets the -t/-T parameters accordingly:\n+        ##- in case of history a symbolic link is used because samtools (view) will generate\n+        ##  the index which might not be possible in the directory containing the fasta file\n+        ##- in case of cached the absolute path is used which allows to read the cram file\n+        ##  without specifying the reference\n+        #if $addref_cond.addref_select == "history":\n+            ln -s \'${addref_cond.ref}\' reference.fa &&\n+            samtools faidx reference.fa &&\n+            #set reffa="reference.fa"\n+            #set reffai="reference.fa.fai"\n+        #elif $addref_cond.addref_select == "cached":\n+            #set reffa=str($addref_cond.ref.fields.path)\n+            #set reffai=str($addref_cond.ref.fields.path)+".fai"\n+        #else\n+            #set reffa=None\n+            #set reffai=None\n+        #end if\n+    ]]></token>\n+\n+    <xml name="optional_reference">\n+        <conditional name="addref_cond">\n+            <param name="addref_select" type="select" label="Use a reference sequence">\n+                <help>@HELP@</help>\n+                <option value="no">No</option>\n+                <option value="history">Use a genome/index from the history</option>\n+                <option value="cached">Use a built-in genome</option>\n+            </param>\n+            <when value="no"/>\n+            <when value="history">\n+                <param name="ref" argument="@ARGUMENT@" type="data" format="fasta,fasta.gz" label="Reference"/>\n+            </when>\n+            <when value="cached">\n+                <param name="ref" argument="@ARGUMENT@" type="select" label="Reference">\n+                    <options from_data_table="fasta_indexes">\n+                        <filter type="data_meta" ref="input" key="dbkey" column="dbkey"/>\n+                    </options>\n+                    <validat'..b'ed="@S1024@">Read is a PCR or optical duplicate</option>\n+        <option value="2048" selected="@S2048@">Alignment is supplementary</option>\n+    </xml>\n+\n+    <!-- region specification macros and tokens for tools that allow the specification\n+         of region by bed file / space separated list of regions -->\n+    <token name="@REGIONS_FILE@"><![CDATA[\n+        #if $cond_region.select_region == \'tab\':\n+            -t \'$cond_region.targetregions\'\n+        #end if\n+    ]]></token>\n+    <token name="@REGIONS_MANUAL@"><![CDATA[\n+        #if $cond_region.select_region == \'text\':\n+            #for $i, $x in enumerate($cond_region.regions_repeat):\n+               \'${x.region}\'\n+            #end for\n+        #end if\n+    ]]></token>\n+    <xml name="regions_macro">\n+        <conditional name="cond_region">\n+            <param name="select_region" type="select" label="Filter by regions" help="restricts output to only those alignments which overlap the specified region(s)">\n+                <option value="no" selected="True">No</option>\n+                <option value="text">Manualy specify regions</option>\n+                <option value="tab">Regions from tabular file</option>\n+            </param>\n+            <when value="no"/>\n+            <when value="text">\n+                <repeat name="regions_repeat" min="1" default="1" title="Regions">\n+                    <param name="region" type="text" label="region" help="format chr:from-to">\n+                        <validator type="regex" message="Required format: CHR[:FROM[-TO]]; where CHR: string containing any character except quotes, whitespace and colon; FROM and TO: any integer">^[^\\s\'\\":]+(:\\d+(-\\d+){0,1}){0,1}$</validator>\n+                    </param>\n+                </repeat>\n+            </when>\n+            <when value="tab">\n+                <param name="targetregions" argument="-t/--target-regions" type="data" format="tabular" label="Target regions file" help="Do stats in these regions only. Tab-delimited file chr,from,to (1-based, inclusive)" />\n+            </when>\n+        </conditional>\n+    </xml>\n+\n+    <xml name="citations">\n+        <citations>\n+            <citation type="bibtex">\n+                @misc{SAM_def,\n+                title={Definition of SAM/BAM format},\n+                url = {https://samtools.github.io/hts-specs/},}\n+            </citation>\n+            <citation type="doi">10.1093/bioinformatics/btp352</citation>\n+            <citation type="doi">10.1093/bioinformatics/btr076</citation>\n+            <citation type="doi">10.1093/bioinformatics/btr509</citation>\n+            <citation type="bibtex">\n+                @misc{Danecek_et_al,\n+                Author={Danecek, P., Schiffels, S., Durbin, R.},\n+                title={Multiallelic calling model in bcftools (-m)},\n+                url = {http://samtools.github.io/bcftools/call-m.pdf},}\n+            </citation>\n+            <citation type="bibtex">\n+                @misc{Durbin_VCQC,\n+                Author={Durbin, R.},\n+                title={Segregation based metric for variant call QC},\n+                url = {http://samtools.github.io/bcftools/rd-SegBias.pdf},}\n+            </citation>\n+            <citation type="bibtex">\n+                @misc{Li_SamMath,\n+                Author={Li, H.},\n+                title={Mathematical Notes on SAMtools Algorithms},\n+                url = {http://www.broadinstitute.org/gatk/media/docs/Samtools.pdf},}\n+            </citation>\n+            <citation type="bibtex">\n+                @misc{SamTools_github,\n+                title={SAMTools GitHub page},\n+                url = {https://github.com/samtools/samtools},}\n+            </citation>\n+        </citations>\n+    </xml>\n+    <xml name="version_command">\n+        <version_command><![CDATA[samtools 2>&1 | grep Version]]></version_command>\n+    </xml>\n+    <xml name="stdio">\n+        <stdio>\n+            <exit_code range="1:" level="fatal" description="Error" />\n+        </stdio>\n+    </xml>\n+</macros>\n'
b
diff -r 000000000000 -r 90f2dd176d80 samtools_coverage.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_coverage.xml Tue Sep 28 15:10:16 2021 +0000
[
@@ -0,0 +1,126 @@
+<tool id="samtools_coverage" name="Samtools coverage" version="@TOOL_VERSION@" profile="@PROFILE@">
+    <description>computes the depth at each position or region</description>
+    <macros>
+        <import>macros.xml</import>
+    </macros>
+    <expand macro="requirements"/>
+    <expand macro="stdio"/>
+    <expand macro="version_command"/>
+    <command><![CDATA[
+        @PREPARE_IDX@
+        samtools coverage
+
+            #if $condition_input.input_pooling == "Yes":
+                ${ ' '.join( [ "'%s'" %  $x for $x in $condition_input.input] ) }
+            #else
+                '$condition_input.input'
+            #end if
+
+            -l $additional_options.min_read_length
+            -q $additional_options.min_mq
+            -Q $additional_options.min_bq
+
+            #set $filter = $additional_options.required_flags
+            @FLAGS@
+            -rf $flags
+
+            #set $filter = $additional_options.skipped_flags
+            @FLAGS@
+            -ff $flags
+
+            #if $condition_histogram.histogram_select == "yes"
+                -m
+                -w $condition_histogram.n_bins
+                #if $condition_histogram.region != "":
+                    -r '$condition_histogram.region'
+                #end if
+            #end if
+
+            -o '$output'
+    ]]></command>
+    <inputs>
+        <conditional name="condition_input">
+            <param name="input_pooling" type="select" label="Are you pooling bam files?" help="Calculate the coverage for multiple bam files" >
+                <option value="No" selected="True">No</option>
+                <option value="Yes">Yes</option>
+            </param>
+            <when value="No" >
+                <param name="input" type="data" format="bam" multiple="true" label="BAM file"/>
+            </when>
+            <when value="Yes">
+                <param name="input" type="data" format="bam" multiple="true" label="BAM files to be pooled"/>
+            </when>
+        </conditional>
+
+        <section name="additional_options" title="Additional Options" expanded="true">
+            <param name="min_read_length" argument="-l" type="integer" min="0" value="0" label="Minimum read length" help="Ignore reads shorter than INT base pairs (Default = 0)" />
+            <param name="min_mq" argument="-q" type="integer" min="0" value="0" label="Minimum mapping quality" help="Minimum mapping quality for an alignment to be used (Default = 0)" />
+            <param name="min_bq" argument="-Q" type="integer" min="0" value="0" label="Minimum base quality" help="Minimum base quality for a base to be considered (Default = 0)" />
+
+            <param name="required_flags" argument="--rf" type="select" multiple="True" label="Require that these flags are set">
+                <expand macro="flag_options" />
+            </param>
+
+            <param name="skipped_flags" argument="--ff" type="select" multiple="True" label="Exclude reads with any of the following flags set">
+                <expand macro="flag_options" s4="true" s256="true" s512="true" s1024="true"/>
+            </param>
+        </section>
+
+        <conditional name="condition_histogram">
+            <param name="histogram_select" argument="-m" type="select" label="Histogram" help="Show histogram instead of tabular output">
+                <option value="no" selected="true">No</option>
+                <option value="yes">Yes</option>
+            </param>
+            <when value="yes">
+                <param name="n_bins" argument="-w" type="integer" min="1" value="100" label="Number of bins in histogram" help="Number of bins in histogram" />
+                <param name="region" argument="-r" type="text" value="" label="Show specified region" help="Show specified region. Format: chr:start-end." />
+            </when>
+            <when value="no" />
+        </conditional>
+    </inputs>
+
+    <outputs>
+        <data name="output" format="tabular" from_work_dir="outfile" label="${tool.name} on ${on_string}" />
+    </outputs>
+
+    <tests>
+        <test>
+            <param name="input" value="test_1.bam" ftype="bam" />
+            <output name="output" file="results_1.tabular" ftype="tabular" />
+        </test>
+        <test>
+            <param name="input" value="test_1.bam" ftype="bam" />
+            <param name="histogram_select" value="yes" />
+            <param name="n_bins" value="50" />
+            <output name="output" file="results_2.txt" ftype="tabular" />
+        </test>
+        <test>
+            <param name="input_pooling" value="Yes" />
+            <param name="input" value="test_1.bam,test_2.bam" ftype="bam" />
+            <param name="histogram_select" value="yes" />
+            <param name="n_bins" value="100" />
+            <output name="output" file="results_3.txt" ftype="tabular" />
+        </test>
+    </tests>
+    <help><![CDATA[
+**What it does**
+
+This tool runs the ``samtools coverage`` command.
+
+Computes the depth at each position or region and draws an ASCII-art histogram or tabulated text. If you select to pool the bam files, then it calculates coverage for the combined files.
+
+The tabulated form uses the following headings:
+
+- rname Reference name / chromosome
+- startpos Start position
+- endpos End position (or sequence length)
+- numreads Number reads aligned to the region (after filtering)
+- covbases Number of covered bases with depth >= 1
+- coverage Proportion of covered bases [0..1]
+- meandepth Mean depth of coverage
+- meanbaseq Mean baseQ in covered region
+- meanmapq Mean mapQ of selected reads
+
+    ]]></help>
+    <expand macro="citations"/>
+</tool>
b
diff -r 000000000000 -r 90f2dd176d80 test-data/test_1.bam
b
Binary file test-data/test_1.bam has changed
b
diff -r 000000000000 -r 90f2dd176d80 test-data/test_2.bam
b
Binary file test-data/test_2.bam has changed