annotate segalign.xml @ 10:ec709ce3d91b draft

planemo upload for repository https://github.com/richard-burhans/galaxytools/tree/main/tools/segalign commit 701d5b6d0b25dbd9d2a8c08234f3aca7ef502131

<tool id="segalign" name="SegAlign" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
    <description>A Scalable GPU System for Pairwise Whole Genome Alignments based on LASTZ's seed-filter-extend paradigm</description>
    <macros>
        <import>macros.xml</import>
        <import>sequence_options.xml</import>
        <import>scoring_options.xml</import>
        <import>seeding_options.xml</import>
        <import>ungapped_extension_options.xml</import>
        <import>gapped_extension_options.xml</import>
        <import>output_options.xml</import>
        <import>segalign_output_options.xml</import>
        <import>system_options.xml</import>
    </macros>
    <expand macro="edam_ontology"/>
    <expand macro="requirements"/>
    <required_files>
        <include path="diagonal_partition.py"/>
        <include path="lastz-cmd.ini"/>
        <include path="package_output.py"/>
        <include path="runner.py"/>
    </required_files>
    <command detect_errors="exit_code"><![CDATA[
##
## https://www.gnu.org/software/coreutils/manual/html_node/nproc-invocation.html
##
## If the OMP_NUM_THREADS or OMP_THREAD_LIMIT environment variables
## are set, then they will determine the minimum and maximum returned
## value respectively.
##
## This is how you tame nproc(1)
##
OMP_THREAD_LIMIT=\${GALAXY_SLOTS:-2}

## Mode ----------------------------------------------------------------

#if str($mode.mode_selector) == "segalign"
    #if str($mode.gapped_extension_options.nogapped) == "false"
        mkdir -p "\$(pwd)/work" && 
        faToTwoBit <(gzip -cdfq '$mode.target') "\$(pwd)/work/ref.2bit" &&
        faToTwoBit <(gzip -cdfq '$mode.query') "\$(pwd)/work/query.2bit" &&
    #end if
    #if str($mode.diagonal_partition_options.diagonal_partition) == "true"
        #set $segalign_mode = "segalign_diagonal_partition"
        ## explicitly calling python to bypass a pulsar bug
        ## https://github.com/galaxyproject/pulsar/issues/341
        python '$__tool_directory__/runner.py'
            --output-type tarball
            --output-file '$segalign_diagonal_partition_output'
            --diagonal-partition
            --num-cpu \${GALAXY_SLOTS:-2}
            --tool_directory '$__tool_directory__'
    #else
        #set $segalign_mode = "segalign"
        python '$__tool_directory__/runner.py'
            --output-type output
            --output-file '$segalign_output'
            --num-cpu \${GALAXY_SLOTS:-2}
            --tool_directory '$__tool_directory__'
    #end if
            '$mode.target'
            '$mode.query'
#else if str($mode.mode_selector) == "segalign_repeat_masker"
    #set $segalign_mode = "segalign_repeat_masker"
        run_segalign_repeat_masker
            '$mode.seq_file'
            --num_cpu \${GALAXY_SLOTS:-2}
#end if

## Sequence Options ----------------------------------------------------

            --strand '$mode.sequence_options.strand_selector'
#if $segalign_mode == "segalign_repeat_masker"
            --neighbor_proportion '$mode.sequence_options.neighbor_proportion'
#end if

## Scoring Options -----------------------------------------------------

#set $scoring_pathname = str($mode.scoring_options.scoring)
#if $scoring_pathname != "None":
            --scoring '$scoring_pathname'
#end if
#if str($mode.scoring_options.ambiguous_selector) != "x"
    #if str($mode.scoring_options.set_ambiguous_params_selector) == "true"
        #set $argument_value = ','.join($mode.scoring_options.ambiguous_selector, $mode.scoring_options.ambiguous_reward, $mode.scoring_options.ambiguous_penalty)
            --ambiguous '$argument_value'
    #else
            --ambiguous '$ambiguous_selector'
    #end if
#end if

## Seeding Options -----------------------------------------------------

#if str($mode.seeding_options.seed.seed_selector) == "custom"
            --seed '$mode.seeding_options.seed.custom_seed'
#else
            --seed '$mode.seeding_options.seed.seed_selector'
#end if
            --step '$mode.seeding_options.step'
#if str($mode.seeding_options.notransition) == "true"
            --notransition
#end if

## Ungapped Extension Options ------------------------------------------

            --xdrop '$mode.ungapped_extension_options.xdrop'
            --hspthresh '$mode.ungapped_extension_options.hspthresh'
#if str($mode.ungapped_extension_options.noentropy) == "true"
            --noentropy
#end if

## Gapped Extension Options --------------------------------------------

#if $segalign_mode != "segalign_repeat_masker"
    #if str($mode.gapped_extension_options.nogapped) == "true"
            --nogapped
    #end if
            --ydrop '$mode.gapped_extension_options.ydrop'
    #if str($mode.gapped_extension_options.gappedthresh) != ""
            --gappedthresh '$mode.gapped_extension_options.gappedthresh'
    #end if
    #if str($mode.gapped_extension_options.notrivial) == "true"
            --notrivial
    #end if
#end if

## Output Options -----------------------------------------------------

#if $segalign_mode != "segalign_repeat_masker"
    #if str($mode.output_options.format.format_selector) == "bam"
            --format '$mode.output_options.format.bam_options'
    #else if str($mode.output_options.format.format_selector) == "general_def"
            --format general-
    #else if str($mode.output_options.format.format_selector) == "general_full"
            --format 'general-:${mode.output_options.format.fields}'
    #else if str($mode.output_options.format.format_selector) == "maf"
            --format '$mode.output_options.format.maf_type'
    #else if str($mode.output_options.format.format_selector) == "blastn"
            --format=BLASTN-
    #else if str($mode.output_options.format.format_selector) == "differences"
            --format=differences
    #end if
    ## todo :: rplot, bam
    ##  --action:target=multiple
    ##  $output_format.rplot
    ##  .if str( $output_format.out.format ) == "bam":
    ##      | samtools sort -@\${GALAXY_SLOTS:-2} -T "\${TMPDIR:-.}" -O bam -o '${output}'
    ##  .else:
    ##      > '${output}'
    ##  .end if
    ##  .if $output_format.rplot:
    ##      &&
    ##      Rscript $r_plot > /dev/null 2>&1
    ##  .end if
#else if $segalign_mode == "segalign_repeat_masker"
            --M '$mode.output_options.M'
            --output '$segalign_repeat_masker_output'
#end if
#if str($mode.output_options.markend) == "true"
            --markend
#end if

## System Options -----------------------------------------------------

            --wga_chunk_size '$mode.system_options.wga_chunk_size'
            --lastz_interval_size '$mode.system_options.lastz_interval_size'
            --seq_block_size '$mode.system_options.seq_block_size'
            --num_gpu '$mode.system_options.num_gpu'
#if str($mode.system_options.debug) == "true"
            --debug
#end if

## -------------------------------------------------------------------

#if $segalign_mode == "segalign_diagonal_partition"
        &&
        python '$__tool_directory__/package_output.py'
#end if

    ]]></command>
    <inputs>
        <conditional name="mode">
            <param name="mode_selector" type="select" label="Choose the mode">
                <option value="segalign" selected="true">SegAlign</option>
                <option value="segalign_repeat_masker">SegAlign repeat masker</option>
            </param>
            <when value="segalign">
                <param name="target" type="data" format="fasta,fasta.gz" label="Target sequence file in FASTA format"/>
                <param name="query" type="data" format="fasta,fasta.gz" label="Query sequence file in FASTA format"/>
                <expand macro="sequence_options"/>
                <expand macro="scoring_options"/>
                <expand macro="seeding_options"/>
                <expand macro="ungapped_extension_options"/>
                <expand macro="gapped_extension_options"/>
                <expand macro="output_options">
                    <expand macro="segalign_output_options"/>
                </expand>
                <expand macro="system_options"/>
                <section name="diagonal_partition_options" expanded="false" title="Diagonal Partition Options">
                    <param argument="--diagonal_partition" type="boolean" value="false" label="Enable diagonal partition optimization"/>
                </section>
            </when>
            <when value="segalign_repeat_masker">
                <param name="seq_file" type="data" format="fasta,fasta.gz" label="Sequence file in FASTA format"/>
                <expand macro="sequence_options">
                    <param argument="--neighbor_proportion" type="float" value="0.2" label="Proportion of neighbouring intervals to align the query interval to"/>
                </expand>
                <expand macro="scoring_options"/>
                <expand macro="seeding_options"/>
                <expand macro="ungapped_extension_options"/>
                <expand macro="output_options">
                    <param argument="--M" type="integer" value="1" max="255" label="report any position that is covered by at least this many alignments; the maximum allowed depth is 255"/>
                </expand>
                <expand macro="system_options"/>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="segalign_output" format="tabular" label="SegAlign on ${on_string}">
            <change_format>
                <when input="mode.output_options.format.format_selector" value="bam" format="bam"/>
                <when input="mode.output_options.format.format_selector" value="maf" format="maf"/>
                <when input="mode.output_options.format.format_selector" value="differences" format="interval"/>
            </change_format>
            <filter>mode['mode_selector'] == 'segalign' and mode['diagonal_partition_options']['diagonal_partition'] is False</filter>
        </data>
        <data name="segalign_diagonal_partition_output" format="tgz" from_work_dir="data_package.tgz" label="SegAlign Diagonal Partition on ${on_string}">
            <filter>mode['mode_selector'] == 'segalign' and mode['diagonal_partition_options']['diagonal_partition'] is True</filter>
        </data>
        <data name="segalign_repeat_masker_output" format="tabular" label="SegAlign Repeat Masker on ${on_string}">
            <filter>mode['mode_selector'] == 'segalign_repeat_masker'</filter>
        </data>
    </outputs>
    <tests>
        <test expect_num_outputs="1" expect_test_failure="true">
            <param name="mode_selector" value="segalign"/>
            <param name="target" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/>
            <param name="query" value="mm39.chr2.chunk.fa.gz" ftype="fasta.gz"/>
            <output name="segalign_output" decompress="true" file="segalign-output.maf.gz" ftype="maf"/>
        </test>
        <test expect_num_outputs="1" expect_test_failure="true">
            <param name="mode_selector" value="segalign_repeat_masker"/>
            <param name="seq_file" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/>
            <output name="segalign_repeat_masker_output" decompress="true" file="segalign-repeat-masker-output.tab.gz" ftype="tabular"/>
        </test>
        <test expect_num_outputs="1" expect_test_failure="true">
            <param name="mode_selector" value="segalign"/>
            <param name="target" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/>
            <param name="query" value="mm39.chr2.chunk.fa.gz" ftype="fasta.gz"/>
            <param name="diagonal_partition" value="true"/>
            <output name="segalign_diagonal_partition_output" ftype="tgz">
                <assert_contents>
                    <has_archive_member path="galaxy/commands.json"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
    SegAlign is a scalable, GPU-accelerated system for computing pairwise WGA. SegAlign is based on the standard seed-filter-extend heuristic, in which the filtering stage dominates the runtime (e.g. 98% for human-mouse WGA), and is accelerated using GPU(s).

    https://github.com/gsneha26
    ]]></help>
    <expand macro="citations"/>
</tool>