comparison hifiasm.xml @ 16:41cea2aa2dce draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/hifiasm commit 7dca92baf6a91f10bced28b883f14d3f4d7630c5

15:60024a528cbd

<tool id="hifiasm" name="Hifiasm" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@">
    <description>haplotype-resolved de novo assembler for PacBio Hifi reads</description>
    <macros>
        <token name="@TOOL_VERSION@">0.19.7</token>
        <token name="@VERSION_SUFFIX@">0</token>
        <token name="@FORMATS@">fasta,fasta.gz,fastq,fastq.gz</token>
        <xml name="reads">
            <param name="reads" type="data" format="@FORMATS@" multiple="true" label="Input reads" />
        </xml>
    </macros>

                ln -s '$read' $inputfile &&
                $ultralong_inputs.append($inputfile)
            #end for
        #end if
        #if str($mode.mode_selector) == 'trio':
            #for idx, read in enumerate($mode.hap1_reads):
                #set $inputfile = 'hap1_input_%d.%s' % ($idx, $read.dataset.extension)
                ln -s '$read' $inputfile &&
                $hap1_inputs.append($inputfile)
            #end for
            #for idx, read in enumerate($mode.hap2_reads):
                #set $inputfile = 'hap2_input_%d.%s' % ($idx, $read.dataset.extension)
                ln -s '$read' $inputfile &&
                $hap2_inputs.append($inputfile)
            #end for
            #set $hap1_filenames = ' '.join($hap1_inputs)
            #set $hap2_filenames = ' '.join($hap2_inputs)
            yak count -k$mode.yak_kmer_length -b$filter_bits -t\${GALAXY_SLOTS:-1} -o hap1.yak $hap1_filenames &&
            yak count -k$mode.yak_kmer_length -b$filter_bits -t\${GALAXY_SLOTS:-1} -o hap2.yak $hap2_filenames &&
        #end if
        hifiasm
        -t \${GALAXY_SLOTS:-1}
        -o output
        -f $filter_bits

            #if $assembly_options.hom_cov
                --hom-cov $assembly_options.hom_cov
            #end if
        #end if
        #if str($mode.mode_selector) == 'trio':
            -1 hap1.yak
            -2 hap2.yak
            -c $mode.max_kmers
            -d $mode.min_kmers
        #end if
        #if str($purge_options.purge_selector) == 'set':
            -l $purge_options.purge_level
            -s $purge_options.similarity_threshold
            -O $purge_options.minimum_overlap

            <when value="standard">
                <expand macro="reads" />
            </when>
            <when value="trio">
                <expand macro="reads" />
                <param name="hap1_reads" type="data" format="fastq,fastq.gz" multiple="true" label="Haplotype 1 reads" />
                <param name="hap2_reads" type="data" format="fastq,fastq.gz" multiple="true" label="Haplotype 2 reads" />
                <param name="max_kmers" argument="-c" type="integer" value="2" label="Lower bound of the binned k-mer's frequency" />
                <param name="min_kmers" argument="-d" type="integer" value="5" label="Upper bound of the binned k-mer's frequency" />
                <param name="yak_kmer_length" type="integer" min="0" max="64" value="31" label="Yak counter k-mer length" />
            </when>
        </conditional>
        <param name="filter_bits" argument="-f" type="integer" min="0" value="37" label="Bits for bloom filter" help="A value of 0 disables the bloom filter" />       
        <conditional name="assembly_options">
            <param name="assembly_selector" type="select" label="Assembly options">

        <!-- TEST 5: Test trio mode -->
        <test expect_num_outputs="5">
            <param name="filter_bits" value="0"/>
            <conditional name="mode">
                <param name="mode_selector" value="trio"/>
                <param name="reads" value="child.fasta.gz"/>
                <param name="hap1_reads" value="paternal.fasta.gz"/>
                <param name="hap2_reads" value="maternal.fasta.gz"/>
                <param name="max_kmers" value="2"/>
                <param name="min_kmers" value="5"/>
            </conditional>
            <assert_command>
                <has_text text="-1 hap1.yak" />

            <param name="filter_bits" value="0" />
            <param name="mode_selector" value="standard" />
            <param name="bins_out" value="yes" />
            <output_collection name="bin_files" type="list" count="3" />
        </test>
    </tests>
    <help><![CDATA[
.. class:: infomark

**HiFiASM - a fast de novo assembler**

16:41cea2aa2dce

<tool id="hifiasm" name="Hifiasm" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@">
    <description>haplotype-resolved de novo assembler for PacBio Hifi reads</description>
    <macros>
        <token name="@TOOL_VERSION@">0.19.7</token>
        <token name="@VERSION_SUFFIX@">1</token>
        <token name="@FORMATS@">fasta,fasta.gz,fastq,fastq.gz</token>
        <xml name="reads">
            <param name="reads" type="data" format="@FORMATS@" multiple="true" label="Input reads" />
        </xml>
    </macros>

                ln -s '$read' $inputfile &&
                $ultralong_inputs.append($inputfile)
            #end for
        #end if
        #if str($mode.mode_selector) == 'trio':
            #if str($mode.trioinput.trio_input_selector) == 'reads':
                #for idx, read in enumerate($mode.trioinput.hap1_reads):
                    #set $inputfile = 'hap1_input_%d.%s' % ($idx, $read.dataset.extension)
                    ln -s '$read' $inputfile &&
                    $hap1_inputs.append($inputfile)
                #end for
                #for idx, read in enumerate($mode.trioinput.hap2_reads):
                    #set $inputfile = 'hap2_input_%d.%s' % ($idx, $read.dataset.extension)
                    ln -s '$read' $inputfile &&
                    $hap2_inputs.append($inputfile)
                #end for
                #set $hap1_filenames = ' '.join($hap1_inputs)
                #set $hap2_filenames = ' '.join($hap2_inputs)
                yak count -k$mode.yak_kmer_length -b$filter_bits -t\${GALAXY_SLOTS:-1} -o hap1.yak $hap1_filenames &&
                yak count -k$mode.yak_kmer_length -b$filter_bits -t\${GALAXY_SLOTS:-1} -o hap2.yak $hap2_filenames &&
            #end if
            #if str($mode.trioinput.trio_input_selector) == 'lists':
                #set $hap1_filenames = $mode.trioinput.hap1_list
                #set $hap2_filenames = $mode.trioinput.hap2_list
            #end if
        #end if
        hifiasm
        -t \${GALAXY_SLOTS:-1}
        -o output
        -f $filter_bits

            #if $assembly_options.hom_cov
                --hom-cov $assembly_options.hom_cov
            #end if
        #end if
        #if str($mode.mode_selector) == 'trio':
            #if str($mode.trioinput.trio_input_selector) == 'reads':
                -1 hap1.yak
                -2 hap2.yak
            #end if
            #if str($mode.trioinput.trio_input_selector) == 'lists':
                -3 $hap1_filenames
                -4 $hap2_filenames
            #end if
            -c $mode.max_kmers
            -d $mode.min_kmers
            $mode.trio_dual
        #end if
        #if str($purge_options.purge_selector) == 'set':
            -l $purge_options.purge_level
            -s $purge_options.similarity_threshold
            -O $purge_options.minimum_overlap

            <when value="standard">
                <expand macro="reads" />
            </when>
            <when value="trio">
                <expand macro="reads" />
                <conditional name="trioinput">
                    <param name="trio_input_selector" type="select" label="What parental information are you using?">
                        <option value="reads">Parental reads (FASTQ files, gzipped or otherwise)</option>
                        <option value="lists">Lists of reads assigned by parent (text files)</option>
                    </param>
                    <when value="reads">
                        <param name="hap1_reads" type="data" format="fastq,fastq.gz" multiple="true" label="Haplotype 1 reads" />
                        <param name="hap2_reads" type="data" format="fastq,fastq.gz" multiple="true" label="Haplotype 2 reads" />
                    </when>
                    <when value="lists">
                        <param name="hap1_list" type="data" format="txt,tabular" label="Haplotype 1 read list" />
                        <param name="hap2_list" type="data" format="txt,tabular" label="Haplotype 2 read list" />
                    </when>
                </conditional>
                <param name="max_kmers" argument="-c" type="integer" value="2" label="Lower bound of the binned k-mer's frequency" />
                <param name="min_kmers" argument="-d" type="integer" value="5" label="Upper bound of the binned k-mer's frequency" />
                <param name="yak_kmer_length" type="integer" min="0" max="64" value="31" label="Yak counter k-mer length" />
                <param name="trio_dual" argument="--trio-dual" type="boolean" truevalue="--trio-dual" falsevalue="" label="Utilize homology information to correct trio-phasing errors" />
            </when>
        </conditional>
        <param name="filter_bits" argument="-f" type="integer" min="0" value="37" label="Bits for bloom filter" help="A value of 0 disables the bloom filter" />       
        <conditional name="assembly_options">
            <param name="assembly_selector" type="select" label="Assembly options">

        <!-- TEST 5: Test trio mode -->
        <test expect_num_outputs="5">
            <param name="filter_bits" value="0"/>
            <conditional name="mode">
                <param name="mode_selector" value="trio"/>
                <param name="trio_input_selector" value="reads"/>
                <conditional name="trioinput">
                    <param name="reads" value="child.fasta.gz"/>
                    <param name="hap1_reads" value="paternal.fasta.gz"/>
                    <param name="hap2_reads" value="maternal.fasta.gz"/>
                </conditional>
                <param name="max_kmers" value="2"/>
                <param name="min_kmers" value="5"/>
            </conditional>
            <assert_command>
                <has_text text="-1 hap1.yak" />

            <param name="filter_bits" value="0" />
            <param name="mode_selector" value="standard" />
            <param name="bins_out" value="yes" />
            <output_collection name="bin_files" type="list" count="3" />
        </test>
        <!-- TEST 15: Test trio LIST mode -->
        <test expect_num_outputs="6">
            <param name="filter_bits" value="0"/>
            <param name="log_out" value="yes"/>
            <conditional name="mode">
                <param name="mode_selector" value="trio"/>
                <param name="reads" value="child.fasta.gz"/>
                <conditional name="trioinput">
                    <param name="trio_input_selector" value="lists"/>
                    <param name="hap1_list" value="maternal.headers.txt"/>
                    <param name="hap2_list" value="paternal.headers.txt"/>
                </conditional>
                <param name="max_kmers" value="2"/>
                <param name="min_kmers" value="5"/>
            </conditional>
            <output name="log_file" ftype="txt">
                <assert_contents>
                    <has_text text="flagged 100 reads, out of 100 lines in file"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
.. class:: infomark

**HiFiASM - a fast de novo assembler**