annotate polypolish.xml @ 2:f355085dd2aa draft default tip

planemo upload for repository https://github.com/mesocentre-clermont-auvergne/galaxy-tools/tree/master/tools/polypolish commit 2a8a85c8b08a62edbd11c34126753192baf6da66

<tool id="polypolish" name="Polypolish" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
    <description>
        Short-read polishing of long-read bacterial genome assemblies
    </description>
    <macros>
        <import>macro.xml</import>
    </macros>
    <expand macro='xrefs'/>
    <expand macro="requirements" />
    <expand macro="version_command" />
    <command detect_errors="aggressive"><![CDATA[
        ln -s '$input.fasta_file' input_data &&
        #*======================================
                    For single SAM
        ======================================*#
        #if $input.sam_data_type.sam_selector == 'single'
            #if $input.sam_data_type.single_sam.is_of_type("unsorted.bam")
                samtools view -@ \$addthreads -h $input.sam_data_type.single_sam > input_sam &&
            #elif $input.sam_data_type.single_sam.ext == 'sam'
                ln -s $input.sam_data_type.single_sam input_sam &&
            #else
                echo "${input.sam_data_type.single_sam} not a sam/bam file"
            #end if
            polypolish input_data input_sam > '$polished_fasta'
        #*======================================
                    For paired SAM
        ======================================*#
        #elif $input.sam_data_type.sam_selector == 'paired'
            #if $input.sam_data_type.R1_sam.is_of_type("unsorted.bam")
                samtools view -@ \$addthreads -h $input.sam_data_type.R1_sam > sample_R1.sam &&
            #elif $input.sam_data_type.R1_sam.ext == 'sam'
                ln -s '$input.sam_data_type.R1_sam' sample_R1.sam &&
            #else
                echo "${input.sam_data_type.single_sam} not a sam/bam file"
            #end if
            #if $input.sam_data_type.R2_sam.is_of_type("unsorted.bam")
                samtools view -@ \$addthreads -h $input.sam_data_type.R2_sam > sample_R2.sam &&
            #elif $input.sam_data_type.R2_sam.ext == 'sam'
                ln -s '$input.sam_data_type.R2_sam' sample_R2.sam &&
            #else
                echo "${input.sam_data_type.single_sam} not a sam/bam file"
            #end if
            #if $input.sam_data_type.insert_filter.filter_select == 'filter'
                polypolish_insert_filter.py --low '$input.sam_data_type.insert_filter.low'
                                            --high '$input.sam_data_type.insert_filter.high'
                                            --in1 sample_R1.sam
                                            --in2 sample_R2.sam
                                            --out1 'filtered_1.sam'
                                            --out2 'filtered_2.sam' &&
                polypolish input_data 'filtered_1.sam' 'filtered_2.sam' >  $polished_fasta
            #else
                polypolish input_data sample_R1.sam sample_R2.sam  >  $polished_fasta
            #end if
        #*======================================
            For multiple single-end SAM
        ======================================*#
        #elif $input.sam_data_type.sam_selector == 'multiple_single'
            mkdir single_collection &&
            #for $value, $single_sam in enumerate($input.sam_data_type.single_collection):
                #if $single_sam.is_of_type("unsorted.bam")
                    samtools view -@ \$addthreads -h $single_sam > 'single_collection/$(single_sam.element_identifier).sam' &&
                #elif $single_sam.ext == 'sam'
                    ln -s $single_sam 'single_collection/$(single_sam.element_identifier).$(single_sam.ext)' &&
                #else
                    echo "${single_sam} is not a sam/bam file"
                #end if
            #end for
            polypolish input_data single_collection/*.sam > '$polished_fasta'
        #*======================================
            For multiple paired-end SAM
        ======================================*#
        #elif $input.sam_data_type.sam_selector == "multiple_paired"
            mkdir paired_collection &&
            #for $value, $paired_sam in enumerate($input.sam_data_type.paired_collection):
                #if $paired_sam.forward.is_of_type("unsorted.bam")
                    samtools view -@ \$addthreads -h $paired_sam.forward > 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam' &&
                #else
                    ln -s '$paired_sam.forward' 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam' &&
                #end if
                #if $paired_sam.reverse.is_of_type("unsorted.bam")
                    samtools view -@ \$addthreads -h $paired_sam.reverse > 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam' &&
                #else
                    ln -s '$paired_sam.reverse' 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam' &&
                #end if
                #if $input.sam_data_type.insert_filter.filter_select == 'filter'
                    polypolish_insert_filter.py --low '$input.sam_data_type.insert_filter.low'
                                                --high '$input.sam_data_type.insert_filter.high'
                                                --in1 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam'
                                                --in2 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam'
                                                --out1 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier)_filtered.sam'
                                                --out2 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier)_filtered.sam' &&
                #end if
            #end for
        #*======================================
                    Filtering option
        ======================================*#
            #if $input.sam_data_type.insert_filter.filter_select == 'filter'
                polypolish input_data paired_collection/*_filtered.sam > '$polished_fasta'
            #else
                polypolish input_data paired_collection/*.sam > '$polished_fasta'
            #end if
        #end if
        #*======================================
            For debug file output
        ======================================*#
        #if $options.debug == 'true'
            --debug $debug_file
        #end if
        #*======================================
            For LOGFILE OUTPUT
        ======================================*#
        #if $options.keep_logfile == 'true'
            | tee '$logfile'
        #end if
        ]]>
    </command>
    <inputs>
        <section name="input" title="Input sequences" expanded="True">
            <param name="fasta_file" type="data" format="fasta" label="Select a draft genome for polishing"
                   help="Fasta sequence to be cleaned using short-reads data"/>
            <conditional name="sam_data_type">
                <param name="sam_selector" type="select" label="Select aligned data to polish" help="Choose number of aligned sam/bam files. Need aligned file with all possible locations in aligner option">
                    <option value="single">Single SAM/BAM file</option>
                    <option value="paired">Paired SAM/BAM files</option>
                    <option value="multiple_single">Multiple single SAM/BAM files</option>
                    <option value="multiple_paired">Multiple paired SAM/BAM files</option>
                </param>
                <when value="single">
                    <param name="single_sam" type="data" format="sam,unsorted.bam" label="Select a SAM/BAM file" help="Specify dataset with only one SAM/BAM file"/>
                </when>
                <when value="paired">
                    <param name="R1_sam" type="data" format="sam,unsorted.bam" label="Select forward SAM/BAM file" help="Specify the forward SAM/BAM files"/>
                    <param name="R2_sam" type="data" format="sam,unsorted.bam" label="Select reverse SAM/BAM file" help="Specify the reverse SAM/BAM files"/>
                    <expand macro="filter_option"/>
                </when>
                <when value="multiple_single">
                    <param name="single_collection" format="sam,unsorted.bam" type="data_collection" collection_type="list" label="Single-end collection" help="Specify a list of single-end dataset"/>
                </when>
                <when value="multiple_paired">
                    <param name="paired_collection" format="sam,unsorted.bam" type="data_collection" collection_type="list:paired" label="Single-end collection" help="Specify a list of single-end dataset"/>
                    <expand macro="filter_option"/>
                </when>
            </conditional>
        </section>
        <section name="options" title="Options" expanded="False">
            <param argument="--min_depth" type="integer" min="0" value="5" label="Minimal depth"
                   help="A base must occur at least this many times in the pileup to be considered valid [default: 5]"/>
            <param argument="--fraction_invalid" type="float" min="0" value="0.2" max="1" label="Minimal invalid fraction"
                   help="A base must make up less than this fraction of the read depth to be considered invalid [default: 0.2]"/>
            <param argument="--max_errors" type="integer" min="0" value="10" label="Number of mismatch/indels to ignore alignments"
                   help="Ignore alignments with more than this many mismatches and indels [default: 10]"/>
            <param argument="--fraction_valid" type="float" min="0" value="0.5" max="1" label="Minimal valid fraction"
                   help="A base must make up at least this fraction of the read depth to be considered valid [default: 0.5"/>
            <param name="keep_logfile" type="boolean" truevalue="true" falsevalue="false" label="Keep log file"/>
            <param argument="--debug" type="boolean" truevalue="true" falsevalue="false" label="Keep per base information file"/>
        </section>
    </inputs>
    <outputs>
        <data name="polished_fasta" format="fasta" label="${tool.name} on ${on_string}: polished fasta"/>
        <data name="debug_file" format="tabular" label="${tool.name} on ${on_string}: Per base informations">
            <filter> options['debug'] == True </filter>
        </data>
        <data name="logfile" format="txt" from_work_dir="output" label="${tool.name} on ${on_string}: log report">
            <filter> options['keep_logfile'] == True </filter>
        </data>
    </outputs>
    <tests>
        <!-- Test_1 with default values and single SAM -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="single"/>
                    <param name="single_sam" value="aligned_test_file/alignement_R1.sam"/>
                </conditional>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_2 with default values and paired SAM -->
        <test expect_num_outputs="2">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="paired"/>
                    <param name="R1_sam" value="aligned_test_file/alignement_R1.sam"/>
                    <param name="R2_sam" value="aligned_test_file/alignement_R2.sam"/>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="true"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
            <output name="debug_file" value="debug_file_test_2.tsv"/>
        </test>
        <!-- Test_3 with default values and single-end multiple SAM -->
        <test expect_num_outputs="2">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="multiple_single"/>
                    <param name="single_collection">
                        <collection type="list">
                            <element name="R1_sam" value="aligned_test_file/alignement_R1.sam" ftype="sam"/>
                            <element name="R1_bis_sam" value="aligned_test_file/alignement_R1_bis.sam" ftype="sam"/>
                            <element name="R1_ter_sam" value="aligned_test_file/alignement_R1_ter.sam" ftype="sam"/>
                        </collection>
                    </param>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="true"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
            <output name="logfile" value="logfile_test_3.log" lines_diff="15"/>
        </test>
        <!-- Test_4 with default values and paired collection SAM -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="multiple_paired"/>
                    <param name="paired_collection">
                        <collection type="list:paired">
                            <element name="paired_1">
                                <collection type="paired">
                                    <element name="forward" value="aligned_test_file/alignement_R1.sam" ftype="sam"/>
                                    <element name="reverse" value="aligned_test_file/alignement_R2.sam" ftype="sam"/>
                                </collection>
                            </element>
                            <element name="paired_2">
                                <collection type="paired">
                                    <element name="forward" value="aligned_test_file/alignement_R1_bis.sam" ftype="sam"/>
                                    <element name="reverse" value="aligned_test_file/alignement_R2_bis.sam" ftype="sam"/>
                                </collection>
                            </element>
                            <element name="paired_3">
                                <collection type="paired">
                                    <element name="forward" value="aligned_test_file/alignement_R1_ter.sam" ftype="sam"/>
                                    <element name="reverse" value="aligned_test_file/alignement_R2_ter.sam" ftype="sam"/>
                                </collection>
                            </element>
                        </collection>
                    </param>
                </conditional>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_5 paired-end without filtering and whitout log file -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="paired"/>
                    <param name="R1_sam" value="aligned_test_file/alignement_R1.sam"/>
                    <param name="R2_sam" value="aligned_test_file/alignement_R2.sam"/>
                </conditional>
            </section>
            <section name="options">
                <conditional name="insert_filter">
                    <param name="filter_select" value="non_filter"/>
                </conditional>
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_6 paired-end with filter, user defined values and whitout log file -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="paired"/>
                    <param name="R1_sam" value="aligned_test_file/alignement_R1.sam"/>
                    <param name="R2_sam" value="aligned_test_file/alignement_R2.sam"/>
                    <conditional name="insert_filter">
                        <param name="filter_select" value="filter"/>
                        <param name="low" value="1"/>
                        <param name="high" value="98.7"/>
                    </conditional>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_7 paired-end with all customized filters -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="paired"/>
                    <param name="R1_sam" value="aligned_test_file/alignement_R1.sam"/>
                    <param name="R2_sam" value="aligned_test_file/alignement_R2.sam"/>
                    <conditional name="insert_filter">
                        <param name="filter_select" value="filter"/>
                        <param name="low" value="1.4"/>
                        <param name="high" value="96.6"/>
                    </conditional>
                </conditional>
            </section>
            <section name="options">
                <param name="min_depth" value="10"/>
                <param name="fraction_invalid" value="0.5"/>
                <param name="max_errors" value="8"/>
                <param name="fraction_valid" value="0.6"/>
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_8 single with bam input -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="single"/>
                    <param name="single_sam" value="aligned_test_file/alignement_R1.bam" ftype="unsorted.bam"/>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_9 paired-end with bam input -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="paired"/>
                    <param name="R1_sam" value="aligned_test_file/alignement_R1.bam" ftype="unsorted.bam"/>
                    <param name="R2_sam" value="aligned_test_file/alignement_R2.bam" ftype="unsorted.bam"/>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_10 single collection with bam input -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="multiple_single"/>
                    <param name="single_collection">
                        <collection type="list">
                            <element name="R1_sam" value="aligned_test_file/alignement_R1.sam" ftype="sam"/>
                            <element name="R1_bis_sam" value="aligned_test_file/alignement_R1_bis.sam" ftype="sam"/>
                            <element name="R1_ter_sam" value="aligned_test_file/alignement_R1_ter.sam" ftype="sam"/>
                        </collection>
                    </param>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_11 paired-end with bam input -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                  <param name="sam_selector" value="multiple_paired"/>
                  <param name="paired_collection">
                      <collection type="list:paired">
                          <element name="paired_1">
                              <collection type="paired">
                                  <element name="forward" value="aligned_test_file/alignement_R1.bam" ftype="unsorted.bam"/>
                                  <element name="reverse" value="aligned_test_file/alignement_R2.bam" ftype="unsorted.bam"/>
                              </collection>
                          </element>
                          <element name="paired_2">
                              <collection type="paired">
                                  <element name="forward" value="aligned_test_file/alignement_R1_bis.bam" ftype="unsorted.bam"/>
                                  <element name="reverse" value="aligned_test_file/alignement_R2_bis.bam" ftype="unsorted.bam"/>
                              </collection>
                          </element>
                          <element name="paired_3">
                              <collection type="paired">
                                  <element name="forward" value="aligned_test_file/alignement_R1_ter.bam" ftype="unsorted.bam"/>
                                  <element name="reverse" value="aligned_test_file/alignement_R2_ter.bam" ftype="unsorted.bam"/>
                              </collection>
                          </element>
                      </collection>
                  </param>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="false"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_12 Test with bam files -->
        <test expect_num_outputs="1">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="single"/>
                    <param name="single_sam" value="aligned_test_file/alignement_R1.bam"/>
                </conditional>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
        </test>
        <!-- Test_13 paired bam -->
        <test expect_num_outputs="2">
            <section name="input">
                <param name="fasta_file" value="contigs.fa"/>
                <conditional name="sam_data_type">
                    <param name="sam_selector" value="paired"/>
                    <param name="R1_sam" value="aligned_test_file/alignement_R1.bam"/>
                    <param name="R2_sam" value="aligned_test_file/alignement_R2.bam"/>
                </conditional>
            </section>
            <section name="options">
                <param name="debug" value="true"/>
                <param name="keep_logfile" value="false"/>
            </section>
            <output name="polished_fasta" value="polished.fasta"/>
            <output name="debug_file" value="debug_file_test_2.tsv"/>
        </test>
    </tests>
  <help><![CDATA[
    **What it does**
    Polypolish is a tool for polishing genome assemblies with short reads.
    Polypolish uses SAM/BAM files where each read has been aligned to all possible locations (not just a single best location).
    This allows it to repair errors in repeat regions that other alignment-based polishers cannot fix.

    **Polypolish pipeline steps**
    1. [Optional] Filter aligned reads
    Exclude some alignments based on their insert size
    This should reduce the number of excessive alignments,
    particularly near the edges of repeat sequences, improving Polypolish's ability to fix errors in those regions.
    2. Clean assembly with filtered reads


    **Inputs**
    Polypolish need SAM/BAM input format obtain from aligner with option to keep all possible location
    Polypolish take on or more assembly as input fasta.
    It need also raw data reads in single or paired-end SAM/BAM format.
    You can use multiple aligned data to polish the same assembly.
    **WARNING It can only work if multiple location information is available in sam/bam files**
    For example using bwa mem to align raw data before use, you need :
    1. To align each read data independantly (also for paired data)
    2. Set the option "Output all alignments for single-ends or unpaired paired-ends" in Select analysis mode>Set input/output options
    This allow multiple ailgnemnt output need to use polypolish
  ]]></help>
    <expand macro="citations"/>
</tool>