Mercurial > repos > iuc > bcftools_plugin_mendelian
changeset 23:eed37e76acd1 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bcftools commit f6efda26965eb73c9107d367fd5ffdf246ed0dbc
| author | iuc |
|---|---|
| date | Tue, 02 Dec 2025 08:14:52 +0000 |
| parents | 8f5962c43868 |
| children | |
| files | bcftools_plugin_mendelian.xml macros.xml |
| diffstat | 2 files changed, 220 insertions(+), 120 deletions(-) [+] |
line wrap: on
line diff
--- a/bcftools_plugin_mendelian.xml Sun Aug 18 10:05:23 2024 +0000 +++ b/bcftools_plugin_mendelian.xml Tue Dec 02 08:14:52 2025 +0000 @@ -1,15 +1,14 @@ <tool name="bcftools @EXECUTABLE@" id="bcftools_plugin_@PLUGIN_ID@" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@"> <description>plugin Count Mendelian consistent / inconsistent genotypes</description> <macros> - <token name="@EXECUTABLE@">mendelian</token> + <token name="@EXECUTABLE@">mendelian2</token> <token name="@PLUGIN_ID@">mendelian</token> <import>macros.xml</import> </macros> - <expand macro="bio_tools" /> + <expand macro="bio_tools" /> <expand macro="requirements" /> <expand macro="version_command" /> <command detect_errors="aggressive"><![CDATA[ -@PREPARE_ENV@ @PREPARE_INPUT_FILE@ #set $section = $sec_restrict @PREPARE_TARGETS_FILE@ @@ -32,25 +31,12 @@ ## Plugin section #set $section = $sec_plugin #if $section.trios.trios_src == 'trio': - --trio "${section.trios.mother},${section.trios.father},${section.trios.child}" + --pfm '${section.trios.num_x}:${section.trios.child},${section.trios.mother},${section.trios.father}' #elif $section.trios.trios_src == 'trio_file': - --trio-file "$section.trios.trio_file" + --ped '$section.trios.trio_file' #end if -#if $section.mode: - #set $modes = str($section.mode).split(',') - #if 'count' in $modes: - --count - #end if - #if 'delete' in $modes: - --delete - #end if - #if '+' in $modes: - --list '+' - #end if - #if 'x' in $modes: - --list x - #end if -#end if +#set $modes = "".join(str($section.mode).split(',')) +-m $modes 2> tmp_stderr > '$output_file' && cat tmp_stderr @@ -59,8 +45,8 @@ <inputs> <expand macro="macro_input" /> <section name="sec_restrict" expanded="false" title="Restrict to"> - <expand macro="macro_restrict" /> - <expand macro="macro_restrict" type="target" label_type="Target" /> + <expand macro="macro_region_restrict" /> + <expand macro="macro_target_restrict" /> <expand macro="macro_include" /> <expand macro="macro_exclude" /> </section> @@ -68,23 +54,28 @@ <!-- trios --> <conditional name="trios"> <param name="trios_src" type="select" label="Sample relationship"> - <option value="trio">trio - mother,father,child</option> - <option value="trio_file">trios file</option> + <option value="trio">trio - mother,father,child (--pfm)</option> + <option value="trio_file">trios file - specify 1 or more trios in PED format (--ped)</option> </param> <when value="trio"> - <param name="mother" type="text" value="" label="name of mother"/> - <param name="father" type="text" value="" label="name of father"/> - <param name="child" type="text" value="" label="name of child"/> + <param name="mother" type="text" value="" label="Mother sample name (as it appears in the input)"/> + <param name="father" type="text" value="" label="Father sample name (as it appears in the input)"/> + <param name="child" type="text" value="" label="Child sample name (as it appears in the input)"/> + <param name="num_x" type="boolean" truevalue="1X" falsevalue="2X" label="Male child?" help="This information is used to judge inheritance patterns of X-chromosomal mutations correctly."/> </when> <when value="trio_file"> - <param name="trio_file" type="data" format="txt" label="Trio File" help="List of trios, one per line" /> + <param name="trio_file" type="data" format="tabular" label="Sample information in PED format" help="This input dataset can describe one or several family trios. See the tool help below for more information on PED format." /> </when> </conditional> - <param name="mode" type="select" label="Action" multiple="true" min="1"> - <option value="count">count the number of consistent sites</option> - <option value="delete">delete inconsistent genotypes (set to './.')</option> - <option value="+">list consistent sites (+)</option> - <option value="x">list inconsistent sites (x)</option> + <param name="mode" type="select" multiple="true" optional="false" label="Output mode" help="Output mode. Multiple modes can be combined and the drop modes take precedence."> + <option value="a" selected="true">add INFO -> MERR annotation with the number of inconsistent trios (a)</option> + <option value="d">delete (i.e., set to "./.") genotypes in inconsistent trios (d)</option> + <option value="e">output sites with at least one erroneous trio (e)</option> + <option value="E">drop sites with at least one erroneous trio (E)</option> + <option value="g">output sites with at least one good (i.e. with non-missing and consistent genotypes) trio (g)</option> + <option value="m">output sites with missing genotypes in at least one trio (m)</option> + <option value="M">drop sites with missing genotypes in at least one trio (M)</option> + <option value="S">drop sites skipped for various reasons when collecting stats (S)</option> </param> </section> <expand macro="macro_select_output_type" /> @@ -96,11 +87,15 @@ <test> <!-- This is just a test on a random file, should have a real test case --> <param name="input_file" ftype="vcf" value="convert.vcf" /> - <param name="trios_src" value="trio" /> - <param name="mother" value="NA00001" /> - <param name="father" value="NA00002" /> - <param name="child" value="NA00006" /> - <param name="mode" value="delete" /> + <section name="sec_plugin"> + <conditional name="trios"> + <param name="trios_src" value="trio" /> + <param name="mother" value="NA00001" /> + <param name="father" value="NA00002" /> + <param name="child" value="NA00006" /> + </conditional> + <param name="mode" value="d" /> + </section> <param name="output_type" value="v" /> <output name="output_file"> <assert_contents> @@ -111,11 +106,15 @@ <test> <!-- This is just a test on a random file, should have a real test case --> <param name="input_file" ftype="vcf" value="convert.vcf" /> - <param name="trios_src" value="trio" /> - <param name="mother" value="NA00001" /> - <param name="father" value="NA00002" /> - <param name="child" value="NA00006" /> - <param name="mode" value="x" /> + <section name="sec_plugin"> + <conditional name="trios"> + <param name="trios_src" value="trio" /> + <param name="mother" value="NA00001" /> + <param name="father" value="NA00002" /> + <param name="child" value="NA00006" /> + </conditional> + <param name="mode" value="e" /> + </section> <param name="output_type" value="v" /> <output name="output_file"> <assert_contents> @@ -124,7 +123,6 @@ </assert_contents> </output> </test> - </tests> <help><![CDATA[ @@ -132,6 +130,43 @@ bcftools @EXECUTABLE@ plugin ===================================== + +Describing sample relationships (family trios) in PED format +------------------------------------------------------------ + +A PED file is simply a tabular text file (columns can be separated by either +spaces or TABs, but not a mixture of the two within the same file) with the +header:: + +#family_id name paternal_id maternal_id sex + +and optional additional columns. The actual column names in the header are not +fixed, but there have to be at least six columns that are interpreted as +detailed next. + +Subsequent lines describe one sample from the VCF input dataset each, where + +- *family_id* is an alphanumeric identifier of a family + + This column has to be present, but is ignored by this plugin + +- *name* is the identifier of the sample described by the line + +- *paternal_id* is the identifier of the sample's father + +- *maternal_id* is the identifier of the sample's mother + +- *sex* is a numeric code (1=male, 2=female) for the sample's sex + +An example minimal PED file describing a single family-trio with a female child could look like this:: + + #family_id name paternal_id maternal_id sex + 0 NA00001 0 0 1 + 0 NA00002 0 0 2 + 0 NA00003 NA00001 NA00002 2 + +where 0 is used as a placeholder for the paternal_id/maternal_id values in the records of the father and mother. + @REGIONS_HELP@ @TARGETS_HELP@ @EXPRESSIONS_HELP@
--- a/macros.xml Sun Aug 18 10:05:23 2024 +0000 +++ b/macros.xml Tue Dec 02 08:14:52 2025 +0000 @@ -1,7 +1,7 @@ <macros> - <token name="@TOOL_VERSION@">1.15.1</token> - <token name="@VERSION_SUFFIX@">4</token> - <token name="@PROFILE@">20.01</token> + <token name="@TOOL_VERSION@">1.22</token> + <token name="@VERSION_SUFFIX@">0</token> + <token name="@PROFILE@">24.1</token> <xml name="bio_tools"> <xrefs> <xref type="bio.tools">bcftools</xref> @@ -10,15 +10,15 @@ <xml name="requirements"> <requirements> <requirement type="package" version="@TOOL_VERSION@">bcftools</requirement> - <requirement type="package" version="1.15.1">htslib</requirement> + <requirement type="package" version="1.22.1">htslib</requirement> <yield /> </requirements> </xml> <xml name="samtools_requirement"> - <requirement type="package" version="1.15.1">samtools</requirement> + <requirement type="package" version="1.22.1">samtools</requirement> </xml> <xml name="matplotlib_requirement"> - <requirement type="package" version="3.5.3">matplotlib</requirement> + <requirement type="package" version="3.10.7">matplotlib</requirement> </xml> <xml name="version_command"> <version_command>bcftools 2>&1 | grep 'Version:'</version_command> @@ -30,16 +30,11 @@ <yield /> </citations> </xml> - <token name="@BCFTOOLS_WIKI@">https://github.com/samtools/bcftools/wiki</token> - <token name="@BCFTOOLS_MANPAGE@">http://samtools.github.io/bcftools/bcftools.html</token> + <token name="@BCFTOOLS_HOWTOS@">https://samtools.github.io/bcftools/howtos/index.html</token> + <token name="@BCFTOOLS_MANPAGE@">https://www.htslib.org/doc/bcftools.html</token> <token name="@THREADS@"> --threads \${GALAXY_SLOTS:-4} </token> - <token name="@PREPARE_ENV@"> -<![CDATA[ -export BCFTOOLS_PLUGINS=`which bcftools | sed 's,bin/bcftools,libexec/bcftools,'`; -]]> - </token> <xml name="macro_input"> <!-- REQUIRES https://github.com/galaxyproject/galaxy/pull/14605/files @@ -294,8 +289,8 @@ help="inverts the query/filtering applied by the targets" /> </xml> - <xml name="macro_restriction_spec" token_type="region" token_label_type="Region"> - <repeat name="@TYPE@s" title="@LABEL_TYPE@ Filter" default="1" min="1"> + <xml name="macro_restriction_spec" tokens="type,label_type"> + <repeat name="@TYPE@_specs" title="@LABEL_TYPE@ Filter" default="1" min="1"> <param name="chrom" type="text" label="@LABEL_TYPE@ chromosome"> <validator type="expression" message="A chromosome identifier is required when specifying a @LABEL_TYPE@ filter">value.strip()</validator> </param> @@ -305,36 +300,85 @@ <param name="stop" type="text" label="@LABEL_TYPE@ end position"> <validator type="expression" message="an integer number is required">not value or value.isdigit()</validator> </param> - <yield /> </repeat> </xml> - <xml name="macro_restrictions_file" token_type="region" token_label_type="Region"> + <xml name="macro_restrictions_file" tokens="type,label_type"> <param name="@TYPE@s_file" type="data" format="tabular" label="@LABEL_TYPE@s File" help="restrict to @LABEL_TYPE@s listed in a file" /> </xml> - <xml name="macro_restrict" token_type="region" token_label_type="Region" > + <xml name="__macro_restrict__" tokens="type,label_select,label_type,pos_is_default,rec_is_default"> <conditional name="@TYPE@s"> - <param name="@TYPE@s_src" type="select" label="@LABEL_TYPE@s"> - <option value="__none__">Do not restrict to @LABEL_TYPE@s</option> - <option value="@TYPE@s">Specify one or more @LABEL_TYPE@(s) directly</option> - <option value="@TYPE@s_file">Operate on @LABEL_TYPE@s specified in a history dataset</option> + <param name="@TYPE@s_src" type="select" label="@LABEL_SELECT@"> + <option value="__none__">Do not restrict to @LABEL_TYPE@s</option> + <option value="@TYPE@s">Specify one or more @LABEL_TYPE@(s) directly</option> + <option value="@TYPE@s_file">Operate on @LABEL_TYPE@s specified in a history dataset</option> + </param> + <when value="__none__"/> + <when value="@TYPE@s"> + <expand macro="macro_restriction_spec" type="@TYPE@" label_type="@LABEL_TYPE@" /> + <yield /> + <param argument="--@TYPE@s-overlap" type="select" label="Consider variant calls with partial overlap with @LABEL_TYPE@(s) if ..." help="Set to 0 if the record has to have POS inside a region; set to 1 if also overlapping records with POS outside a region should be included (this includes indels with POS at the end of a region, which are technically outside the region); or set to 2 to include only true overlapping variation (compare the full VCF representation 'TA>T-' vs the true sequence variation 'A>-')."> + <option value="0" selected="@POS_IS_DEFAULT@">0: the variant's POS falls in the region</option> + <option value="1" selected="@REC_IS_DEFAULT@">1: any part of the variant's representation overlaps the region</option> + <option value="2">2: the variant's actual sequence change falls in the region</option> + </param> + </when> + <when value="@TYPE@s_file"> + <expand macro="macro_restrictions_file" type="@TYPE@" label_type="@LABEL_TYPE@" /> + <yield /> + <param argument="--@TYPE@s-overlap" type="select" label="@LABEL_TYPE@ overlap" help="Include if POS in the region (0), record overlaps (1), variant overlaps (2)"> + <option value="0" selected="@POS_IS_DEFAULT@">0: POS in the region</option> + <option value="1" selected="@REC_IS_DEFAULT@">1: Record overlaps</option> + <option value="2">2: Variant overlaps</option> </param> - <when value="__none__"/> - <when value="@TYPE@s"> - <expand macro="macro_restriction_spec" type="@TYPE@" label_type="@LABEL_TYPE@" /> - <yield /> - </when> - <when value="@TYPE@s_file"> - <expand macro="macro_restrictions_file" type="@TYPE@" label_type="@LABEL_TYPE@" /> - <yield /> - </when> + </when> + </conditional> + </xml> + + <xml name="__macro_restrict_simple__" tokens="type,label_select,label_type"> + <conditional name="@TYPE@s"> + <param name="@TYPE@s_src" type="select" label="@LABEL_SELECT@"> + <option value="__none__">Do not restrict to @LABEL_TYPE@s</option> + <option value="@TYPE@s">Specify one or more @LABEL_TYPE@(s) directly</option> + <option value="@TYPE@s_file">Operate on @LABEL_TYPE@s specified in a history dataset</option> + </param> + <when value="__none__"/> + <when value="@TYPE@s"> + <expand macro="macro_restriction_spec" type="@TYPE@" label_type="@LABEL_TYPE@" /> + <yield /> + </when> + <when value="@TYPE@s_file"> + <expand macro="macro_restrictions_file" type="@TYPE@" label_type="@LABEL_TYPE@" /> + <yield /> + </when> </conditional> - <param argument="--@TYPE@s-overlap" type="select" optional="true" label="@LABEL_TYPE@ overlap" help="Include if POS in the region (0), record overlaps (1), variant overlaps (2)"> - <option value="0">0: POS in the region</option> - <option value="1">1: Record overlaps</option> - <option value="2">2: Variant overlaps</option> - </param> + </xml> + + <xml name="macro_region_restrict" token_label_select="Regions"> + <expand macro="__macro_restrict__" type="region" label_select="@LABEL_SELECT@" label_type="Region" pos_is_default="false" rec_is_default="true" /> + </xml> + + <xml name="macro_region_restrict_simple" token_label_select="Regions"> + <!-- To be used with subcommands that don't support the 'regions-overlap' option. Currently only mpileup? --> + <expand macro="__macro_restrict_simple__" type="region" label_select="@LABEL_SELECT@" label_type="Region" /> + </xml> + + <xml name="macro_target_restrict"> + <expand macro="__macro_restrict__" type="target" label_select="Target" label_type="Target" pos_is_default="true" rec_is_default="false"> + <expand macro="macro_invert_targets" /> + </expand> + </xml> + + <xml name="macro_target_restrict_simple"> + <!-- To be used with subcommands that don't support the 'regions-overlap' option. Currently only mpileup and call? --> + <expand macro="__macro_restrict_simple__" type="target" label_select="Target" label_type="Target"> + <expand macro="macro_invert_targets" /> + </expand> + </xml> + + <xml name="macro_mask_restrict"> + <expand macro="__macro_restrict__" type="mask" label_select="Mask" label_type="Mask" pos_is_default="false" rec_is_default="true" /> </xml> <token name="@PARSE_INTERVALS@"> @@ -345,7 +389,7 @@ #set $start = str($i.start).strip() #set $stop = str($i.stop).strip() #if $start or $stop: - $components.append($chrom + ':' + ($start or '0') + '-' + $stop) + $components.append($chrom + ':' + ($start or '1') + '-' + $stop) #else: $components.append($chrom) #end if @@ -357,18 +401,14 @@ <token name="@MASK@"> <![CDATA[ #if $section.conditional_soft_filter.selector == 'enabled' and $section.conditional_soft_filter.soft_filter - #if $section.conditional_soft_filter.masks.masks_src == 'regions': - #set $intervals = $section.conditional_soft_filter.masks.masks - @PARSE_INTERVALS@ - --mask '$intervals_spec' - #elif $section.conditional_soft_filter.masks.masks_src == 'masks_file' and $section.conditional_soft_filter.masks.masks_file: - #if $masks_path is not None: - --mask-file '$masks_path' - #else: + #if $section.conditional_soft_filter.masks.masks_src != '__none__': + #if $section.conditional_soft_filter.masks.masks_src == 'masks': + #set $intervals = $section.conditional_soft_filter.masks.mask_specs + @PARSE_INTERVALS@ + --mask '$intervals_spec' + #elif $section.conditional_soft_filter.masks.masks_src == 'masks_file': --mask-file '$section.conditional_soft_filter.masks.masks_file' #end if - #end if - #if $section.conditional_soft_filter.masks_overlap --mask-overlap $section.conditional_soft_filter.masks_overlap #end if #end if @@ -378,37 +418,42 @@ <token name="@REGIONS@"> <![CDATA[ -#if $section.regions.regions_src == 'regions': - #set $intervals = $section.regions.regions - @PARSE_INTERVALS@ - --regions '$intervals_spec' -#elif $section.regions.regions_src == 'regions_file' and $section.regions.regions_file: - #if $regions_path is not None: - --regions-file '$regions_path' - #else: - --regions-file '$section.regions.regions_file' +#if $section.regions.regions_src != '__none__': + #if $section.regions.regions_src == 'regions': + #set $intervals = $section.regions.region_specs + @PARSE_INTERVALS@ + --regions '$intervals_spec' + #elif $section.regions.regions_src == 'regions_file': + #if $regions_path is not None: + --regions-file '$regions_path' + #else: + --regions-file '$section.regions.regions_file' + #end if + #end if + #if 'regions_overlap' in $section.regions: + --regions-overlap $section.regions.regions_overlap #end if #end if -#if $section.regions_overlap - --regions-overlap $section.regions_overlap -#end if - ]]> </token> <token name="@TARGETS@"> <![CDATA[ -#if $targets_path: - --targets-file "${section.targets.invert_targets_file}${targets_path}" -#elif $section.targets.targets_src == 'targets': - #set $intervals = $section.targets.targets - @PARSE_INTERVALS@ - --targets '${section.targets.invert_targets_file}$intervals_spec' -#elif $section.targets.targets_src == 'targets_file' and $section.targets.targets_file: - --targets-file "${section.targets.invert_targets_file}${section.targets.targets_file}" -#end if -#if $section.targets_overlap - --targets-overlap $section.targets_overlap +#if $section.targets.targets_src != '__none__': + #if $section.targets.targets_src == 'targets': + #set $intervals = $section.targets.target_specs + @PARSE_INTERVALS@ + --targets '${section.targets.invert_targets_file}$intervals_spec' + #elif $section.targets.targets_src == 'targets_file': + #if $targets_path: + --targets-file '${section.targets.invert_targets_file}${targets_path}' + #else: + --targets-file '${section.targets.invert_targets_file}${section.targets.targets_file}' + #end if + #end if + #if 'targets_overlap' in $section.targets: + --targets-overlap $section.targets.targets_overlap + #end if #end if ]]> </token> @@ -417,7 +462,7 @@ <![CDATA[ #set $regions_path = None #if 'regions' in $section - #if $section.regions.regions_src == 'regions_file' and $section.regions.regions_file: + #if $section.regions.regions_src == 'regions_file': #if $section.regions.regions_file.ext.startswith('bed'): #set $regions_path = 'regions_file.bed' ln -s '$section.regions.regions_file' $regions_path && @@ -461,11 +506,31 @@ </param> <param name="invert_samples" type="boolean" truevalue="^" falsevalue="" checked="false" label="Invert Samples" help="Inverts the query/filtering applied by Samples (adds "^" prefix to exclude)" /> - <param argument="--samples_file" type="data" format="tabular" optional="true" label="Samples file" - help="File of samples to include" /> + <param argument="--samples_file" type="data" format="tabular" optional="true" label="Samples file" help="File of samples to include" /> <param name="invert_samples_file" type="boolean" truevalue="^" falsevalue="" checked="false" label="Invert Samples file" help="inverts the query/filtering applied by Samples file" /> </xml> + + <xml name="macro_samples_enhanced" token_all_samples="" token_file_help="File listing samples to include one per line."> + <conditional name="specify_samples"> + <param name="how" type="select" label="How do you want to select samples?"> + <option value="-s">Specify samples manually</option> + <option value="-S">Provide sample file</option> + <option value="@ALL_SAMPLES@" selected="true">No selection; just use all samples found in input</option> + </param> + <when value="-s"> + <param argument="--samples" name="samples_spec" type="text" optional="false" label="Sample(s)" help="Specify a single sample by name or a comma-separated list of sample names."> + <validator type="expression" message="Please specify a single sample by name or a comma-separated list of sample names">value.strip() and value[0] not in "^," and ",," not in value</validator> + </param> + <param name="invert_samples" type="boolean" truevalue="^" falsevalue="" label="Exclude these samples instead of including them?" help="Equivalent to prefixing the list of samples with "^" on the command line."/> + </when> + <when value="-S"> + <param argument="--samples_file" type="data" format="tabular" optional="@IS_OPTIONAL@" label="Samples file" help="@FILE_HELP@" /> + <param name="invert_samples" type="boolean" truevalue="^" falsevalue="" label="Exclude the samples listed in the file instead of including them?" help="Equivalent to prefixing the file with "^" on the command line."/> + </when> + <when value="@ALL_SAMPLES@" /> + </conditional> + </xml> <token name="@SAMPLES@"> #set $samples_defined = False #if str($section.samples) != '': @@ -478,8 +543,8 @@ #end if </token> - <xml name="macro_sample"> - <param name="sample" type="text" optional="true" label="Sample" help="Apply variants of the given sample" /> + <xml name="macro_sample" token_help="Apply variants of the given sample"> + <param name="sample" type="text" optional="true" label="Sample" help="@HELP@" /> </xml> <token name="@SAMPLE@"> #if $section.sample: