<tool id="b2b_summarize_assembly" name="Summarize assembly" version="0.02">

    <description>Calculate and print basic assembly summary stats</description>

    <!-- ***************************************************************** -->

    <requirements>
    <!--
        <requirement type="package" version="0.003">b2b-utils</requirement>
    -->
        <requirement type="package" version="0.006007">perl-biox-seq</requirement>
    </requirements>

    <!-- ***************************************************************** -->

    <version_command>summarize_assembly --version | perl -wnE'print "$1\n" for /summarize_assembly v(.+)/g'</version_command>

    <!-- ***************************************************************** -->

    <command detect_errors="aggressive">
    <![CDATA[

    summarize_assembly
    --fasta $input1
    --cutoffs $cutoffs
    #if $split_N:
	--split_N
    #end if
    #if $strip_N:
	--strip_N
    #end if

    > $report
    ;

    ]]>
    </command>

    <!-- ***************************************************************** -->

    <inputs>
	    <param format="fasta" name="input1" type="data" label="Assembly in FASTA format" />
	    <param name="cutoffs" type="text" size="60" value="50 90" label="Nx cutoffs to calculate (space-separated)">
		   <sanitizer invalid_char="">
		      <valid initial="string.digits"><add value=" " /></valid>
		   </sanitizer>
	    </param>
	    <param name="strip_N" type="boolean" label="Strip Ns prior to calculations" />
	    <param name="split_N" type="boolean" label="Split scaffolds on Ns (e.g. report contig-level summary)" />
    </inputs>

    <!-- ***************************************************************** -->

    <outputs>
       <data format="txt" name="report" label="${tool.name} on ${on_string}" />
    </outputs>

    <!-- ***************************************************************** -->

    <tests>
        <test>
            <param name="input1" value="assembly.fa" ftype="fasta" />
            <param name="cutoffs" value="50 90 95" />
            <output name="report" file="summarize.txt" compare="diff" />
        </test>
        <test>
            <param name="input1" value="assembly.fa" ftype="fasta" />
            <param name="cutoffs" value="50 90 95" />
            <param name="strip_N" value="True" />
            <output name="report" file="summarize.strip.txt" compare="diff" />
        </test>
        <test>
            <param name="input1" value="assembly.fa" ftype="fasta" />
            <param name="cutoffs" value="50 90 95" />
            <param name="split_N" value="True" />
            <output name="report" file="summarize.split.txt" compare="diff" />
        </test>
    </tests>

    <!-- ***************************************************************** -->

    <help>

    summarize_assembly is a utility from b2b-utils. It takes a FASTA file and
    optionally a list of cutoff values as input and prints out summary
    information about the contigs/scaffolds contained in the file. You can, of
    course, supply a FASTA file of any sort of nucleic acid sequences, but the
    summary information makes most sense for contigs from genomic sequencing
    assemblies.

    </help>

    <!-- ***************************************************************** -->

    <citations>
    </citations>

</tool>