annotate yahs.xml @ 4:425e63adcc5d draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/yahs commit be65e1846c9d4d63e80c01625e6e5f0d613948fe

<tool id="yahs" name="YAHS" version="@VERSION@+galaxy@VERSION_SUFFIX@" profile="20.01">
    <description>yet another HI-C scaffolding tool</description>
    <macros>
        <token name="@VERSION@">1.2a.2</token>
        <token name="@VERSION_SUFFIX@">1</token>
    </macros>
    <requirements>
        <requirement type="package" version="@VERSION@">yahs</requirement>
        <requirement type="package" version="1.11">samtools</requirement>
        <requirement type="package" version="3.9">python</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        #if $function.function_select == "yahs":
            ln -s '$function.fasta' input.fasta &&
            #if $function.bfile.ext == "bam":
                ln -s '$function.bfile' input.bam &&
            #else if $function.bfile.ext == "bed":
                ln -s '$function.bfile' input.bed &&
            #end if
            #if $function.agp:
                ln -s '$function.agp' input.agp &&
            #end if
            samtools faidx input.fasta &&
            mkdir initial_break agp_out agp_break final_outs &&
            yahs --no-mem-check input.fasta 
            #if $function.bfile.ext == "bam":
                input.bam
            #else if $function.bfile.ext == "bed":
                input.bed
            #end if
            #if $agp:
                -a input.agp
            #end if
            #if $function.res:
                -r $function.res
            #end if
            #if $function.enzyme_conditional.enzyme_options == 'preconfigured':
                #if $function.enzyme_conditional.preconfigured_enzymes == 'dovetail'
                    -e 'GATC'
                #else if $function.enzyme_conditional.preconfigured_enzymes == 'arima1'
                    -e 'GATC,GANTC'
                #else if $function.enzyme_conditional.preconfigured_enzymes == 'arima2'
                    -e 'GATC,GANTC,CTNAG,TTAA'
                #else if $function.enzyme_conditional.preconfigured_enzymes == 'omnic'

                #end if
            #else if $function.enzyme_conditional.enzyme_options == 'specific':
                -e '${function.enzyme_conditional.manual_enzyme}'
            #end if
            #if $function.length:
                -l $function.length
            #end if
            #if $function.quality:
                -q $function.quality
            #end if
            $function.no_contig_ec
            $function.no_scaffold_ec
            -o yahs_out
            #if $log_out
                2> output.log
            #end if
            && ls -lah &&
            python '$__tool_directory__/move_files.py'
            && ls -lah ./*
        #else:
            ln -s '$function.fasta' input.fasta &&
            ln -s '$function.agp' input.agp &&
            agp_to_fasta input.agp input.fasta
            #if $function.length:
                -l $function.length
            #end if
            -o output.fasta
            #if $log_out:
                2> output.log
            #end if
        #end if
    ]]></command>
    <inputs>
        <conditional name="function">
            <param name="function_select" type="select" label="Which function would you like to run?">
                <option value="yahs">YAHS</option>
                <option value="agp_to_fasta">AGP to fasta</option>
            </param>
            <when value="yahs">
                <param name="fasta" type="data" format="fasta" label="Input contig sequences"/>
                <param name="bfile" type="data" format="bam,bed" label="Alignment file of Hi-C reads to contigs"/>
                <param name="agp" argument="-a" type="data" format="agp" optional="true" label="Input AGP file (for rescaffolding)"
                    help="You can specify a AGP format file to ask YaHS to do scaffolding with the scaffolds in the AGP file as the start point"/>
                <param name="res" argument="-r" type="text" label="Resolutions" optional="true" 
                    help="Comma separated, ascending list of range of resolutions with no spaces. Ex. 50000,100000,200000,500000,1000000,2000000,5000000. By default and the upper limit is automatically adjusted with the genome size"/>
                <conditional name="enzyme_conditional">
                    <param name="enzyme_options" type="select" label="Restriction enzyme used in Hi-C experiment" help="Hi-C experiments can use different restriction enzymes.
                        The enzyme frequency in contigs is used to normalize the Hi-C interaction frequency. Note that you need to specify the actual 
                        sequence of the cutting site for a restriction enzyme and not the enzyme name. You can also specify DNASE as an enzyme if you 
                        use an enzyme-free prep, e.g. Omin-C.">
                        <option value="not_specified">Not specified</option>
                        <option value="preconfigured">Preconfigured restriction enzymes</option>
                        <option value="specific">Enter a specific sequence</option>
                    </param>
                    <when value="not_specified"/>
                    <when value="preconfigured">
                        <param name="preconfigured_enzymes" argument="-e" type="select" label="Preconfigured enzymes">
                            <option value="dovetail">Dovetail Chicago, Dovetail Hi-C or Phase: GATC</option>
                            <option value="arima1">Arima Hi-C 1.0: GATC, GANTC</option>
                            <option value="arima2">Arima Hi-C 2.0: GATC, GANTC, CTNAG, TTAA</option>
                            <option value="omnic">Dovetail Omni-C: enzyme-free prep</option>
                        </param>
                    </when>
                    <when value="specific">
                        <param name="manual_enzyme" argument="-e" type="text" label="Restriction enzyme sequence(s)"
                            help="Restriction enzyme sequence. If multiple were used, include all as a comma separated list without spaces (ex. 'GATC,AAGCTT').">
                            <validator type="expression" message="Only alphabetical letters and the comma can be used in to define restriction enzym sequences.">value.replace(',', '').isalpha()</validator>
                        </param>
                    </when>
                </conditional>
                <param name="length" argument="-l" type="integer" label="Minimum contig length included for scaffolding" min="1" optional="true"/>
                <param name="quality" argument="-q" type="integer" label="Minimum read mapping quality (for BAM input only)" min="1" optional="true"/>
                <param argument="--no-contig-ec" type="boolean" label="Skip assembly error correction step" truevalue="--no-contig-ec" falsevalue="" help="Automatically active if AGP output is set"/>
                <param argument="--no-scaffold-ec" type="boolean" label="Skip assembly/scaffolding error check each round" truevalue="--no-scaffold-ec" falsevalue="" help="Will also prevent any break.agp output files"/>
            </when>
            <when value="agp_to_fasta">
                <param name="agp" type="data" format="agp" label="Input AGP file"/>
                <param name="fasta" type="data" format="fasta" label="Contig fasta file"/>
                <param name="length" type='integer' label="Output fasta line length" value="60" min="1"/>
            </when>
        </conditional>
        <param name="log_out" type="boolean" label="Output log file?" truevalue="yes" falsevalue="no"/>
    </inputs>
    <outputs>
        <!-- INITIAL ROUNDS -->
        <collection name="initial_agp_break" type="list" label="${tool.name} on ${on_string}: AGP initial break files">
            <filter>function['function_select'] == "yahs"</filter>
            <discover_datasets pattern="__name_and_ext__" directory="initial_break" ext="agp"/>
        </collection>
        <collection name="agp_break" type="list" label="${tool.name} on ${on_string}: AGP break files">
            <filter> function['function_select'] == "yahs"</filter>
            <discover_datasets pattern="__name_and_ext__" directory="agp_break" ext="agp"/>
        </collection>
        <collection name="agp_out" type="list" label="${tool.name} on ${on_string}: AGP scaffolding results files">
            <filter>function['function_select'] == "yahs"</filter>
            <discover_datasets pattern="__name_and_ext__" directory="agp_out" ext="agp"/>
        </collection>
        <!-- FINAL FILES -->
        <data format="agp" name="final_agp_out" label="${tool.name} on ${on_string}: Final scaffolds agp output" from_work_dir="final_outs/yahs_out_scaffolds_final.agp">
            <filter>function['function_select'] == "yahs"</filter>
        </data>
        <data format="fasta" name="final_fasta_out" label="${tool.name} on ${on_string}: Final scaffolds fasta output" from_work_dir="final_outs/yahs_out_scaffolds_final.fa">
            <filter>function['function_select'] == "yahs"</filter>
        </data>
        <data format="fasta" name="fasta_from_agp" label="${tool.name} on ${on_string}: agp to fasta" from_work_dir="output.fasta">
            <filter>function['function_select'] == "agp_to_fasta"</filter>
        </data>
        <data format="txt" name="log_file" from_work_dir="output.log" label="${tool.name} ${on_string}, log file">
            <filter>log_out</filter>
        </data>
    </outputs>
    <tests>
        <!-- TEST 1 -->
        <test expect_num_outputs="5">
            <conditional name="function">
                <param name="function_select" value="yahs"/>
                <param name="fasta" value="test.fasta"/>
                <param name="bfile" value="test.bed"/>
                <param name="agp" value="test.agp"/>
                <param name="res" value="50000,100000,150000,2000000,1000000"/>
            </conditional>
            <output name="final_agp_out" file="test_01_scaffolds_final.agp" ftype="agp"/>
            <output name="final_fasta_out" file="test_01_scaffolds_final.fa" ftype="fasta"/>
            <!-- COMMAND:   yahs test.fasta test.bed -r 50000,100000,150000,2000000,1000000 -a test.agp -o test_1 -->
        </test>
        <!-- TEST 2 -->
        <test expect_num_outputs="5">
            <conditional name="function">
                <param name="function_select" value="yahs"/>
                <param name="fasta" value="test.fasta"/>
                <param name="bfile" value="test.bed"/>
                <param name="no_contig_ec" value="--no-contig-ec"/>
                <param name="no_scaffold_ec" value="--no-scaffold-ec"/>
            </conditional>
            <output name="final_agp_out" file="test_02_scaffolds_final.agp" ftype="agp"/>
            <output name="final_fasta_out" file="test_02_scaffolds_final.fa" ftype="fasta"/>
            <output_collection name="agp_break">
                <element name="yahs_out_no_break" file="test_02_no_break.agp" ftype="agp"/>
            </output_collection>
            <!-- COMMAND:  yahs test.fasta test.bed -\-no-contig-ec -\-no-scaffold-ec -o test_2 -->
        </test>
        <!-- TEST 3 -->
        <test expect_num_outputs="6">
            <conditional name="function">
                <param name="function_select" value="yahs"/>
                <param name="fasta" value="test2.fasta"/>
                <param name="bfile" value="test2.bam"/>
                <param name="res" value="1000,2000,5000,10000,20000,50000,100000,200000,500000"/>
                <conditional name="enzyme_conditional">
                    <param name="enzyme_options" value="not_specified"/>
                </conditional>
            </conditional>
            <param name="log_out" value="yes"/>
	    <output name="log_file" ftype="txt">
            <assert_contents>
                <has_text text="[I::dump_links_from_bam_file] dumped 6399 read pairs from 17675 records: 6297 intra links + 102 inter links" />
            </assert_contents>
	    </output>
            <!-- COMMAND:  yahs test.fasta test.bam -r 1000,2000,5000,10000,20000,50000,100000,200000,500000 -o test_3 -->
        </test>
        <!-- TEST 4 -->
        <test expect_num_outputs="5">
            <conditional name="function">
                <param name="function_select" value="yahs"/>
                <param name="fasta" value="test2.fasta"/>
                <param name="bfile" value="test2.bed"/>
                <param name="qual" value="10"/>
                <param name="length" value="20"/>
                <conditional name="enzyme_conditional">
                    <param name="enzyme_options" value="not_specified"/>
                </conditional>
            </conditional>
            <output name="final_agp_out" file="test_04_scaffolds_final.agp" ftype="agp"/>
            <output name="final_fasta_out" file="test_04_scaffolds_final.fa" ftype="fasta"/>
            <output_collection name="initial_agp_break">
                <element name="yahs_out_inital_break_01" file="test_04_initial_break.agp" ftype="agp"/>
            </output_collection>
        </test>
        <!-- TEST 5 -->
        <test expect_num_outputs="1">
            <conditional name="function">
                <param name="function_select" value="agp_to_fasta"/>
                <param name="fasta" value="test.fasta"/>
                <param name="agp" value="test.agp"/>
                <param name="length" value="20"/>
            </conditional>
            <output name="fasta_from_agp" file="test_05.fasta" ftype="fasta"/>
            <!-- COMMAND: agp_to_fasta test.fasta test.agp -l 20 -o test_4 -->
        </test>
        <!-- TEST 6 -->
        <test expect_num_outputs="6">
            <conditional name="function">
                <param name="function_select" value="yahs"/>
                <param name="fasta" value="test.fasta"/>
                <param name="bfile" value="test.bed"/>
                <param name="agp" value="test.agp"/>
                <param name="res" value="50000,100000,150000,2000000,1000000"/>
            </conditional>
            <param name="log_out" value="yes"/>
            <output name="final_agp_out" file="test_01_scaffolds_final.agp" ftype="agp"/>
            <output name="final_fasta_out" file="test_01_scaffolds_final.fa" ftype="fasta"/>
            <output name="log_file" ftype="txt">
                <assert_contents>
                    <has_text text="[I::main] CMD: yahs"/>
                </assert_contents>
            </output>
            <!-- COMMAND:   yahs test.fasta test.bed -r 50000,100000,150000,2000000,1000000 -a test.agp -o test_1 -->
        </test>
        <!-- TEST 7: omnic prep -->
        <test expect_num_outputs="6">
            <conditional name="function">
                <param name="function_select" value="yahs"/>
                <param name="fasta" value="test.fasta"/>
                <param name="bfile" value="test.bed"/>
            </conditional>
            <conditional name="enzyme_conditional">
                <param name="enzyme_options" value="preconfigured"/>
                <param name="preconfigured_enzymes" value="omnic"/>
            </conditional>
            <param name="log_out" value="yes"/>
            <output name="log_file" ftype="txt">
                <assert_contents>
                    <not_has_text text="-e"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
        YaHS is scaffolding tool using Hi-C data. It relies on a new algothrim for contig joining detection which considers the topological distribution of Hi-C signals aiming to distingush real interaction signals from mapping nosies. YaHS has been tested in a wide range of genome assemblies. Compared to other Hi-C scaffolding tools, it usually generates more contiguous scaffolds - especially with a higher N90 and L90 statistics. It is also super fast - takes less than 5 minutes to reconstruct the human genome from an assembly of 5,483 contigs with ~45X Hi-C data.
    ]]></help>
    <citations>
        <citation type="doi">10.5281/zenodo.5848772</citation>
    </citations>
</tool>