<tool id="schicexplorer_schicdemultiplex" name="@BINARY@" version="@TOOL_VERSION@.1" profile="@PROFILE@">
    <description>demultiplexes Nagano 2017 raw fastq files</description>
    <macros>
        <token name="@BINARY@">scHicDemultiplex</token>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements" />
    <command detect_errors="exit_code"><![CDATA[
        #import re
        mkdir outputFiles &&
        #set fastqFile = @ESCAPE_IDENTIFIER_FASTQ@
        ln -s '$fastq' '$fastqFile' &&
        @BINARY@

        --fastq $fastqFile
        --barcodeFile '$barcodeFile'
        --srrToSampleFile '$srrToSample'
        --outputFolder outputFiles
        --bufferSize $bufferSize
        --threads @THREADS@

        && cd outputFiles
        && mkdir forward
        && mkdir reverse
        && mv *R1.fastq.gz forward
        && mv *R2.fastq.gz reverse

    ]]></command>
    <inputs>
        
        <param name='fastq' type="data" format="fastq, fastq.gz" label="Raw fastq file from Nagano 2017."/>
        <param name='barcodeFile' type="data" format="txt" label="barcode file" help="Use GSE94489_README.txt file "/>
        <param name='srrToSample' type="data" format="txt" label="SRR to Sample mapping file" help="The mappings from SRR number to sample id as given in the barcode file."/>
        <param name="bufferSize" type="integer" value="20000000"  label="Buffer size" help="Number of lines to buffer in memory, if full, write the data to disk." />   

    </inputs>
    <outputs>
        <collection name="outputFastqsForward" type="list" label="Demultiplexed fastq files: Forward">
            <discover_datasets pattern="__name__" format="fastqsanger.gz" directory="outputFiles/forward" />
        </collection>
        <collection name="outputFastqsReverse" type="list" label="Demultiplexed fastq files: Reverse">
            <discover_datasets pattern="__name__" format="fastqsanger.gz" directory="outputFiles/reverse" />
        </collection>
    </outputs>
    <tests>
        <test>
            <param name='fastq' value='scHicDemultiplex/SRR5229025.fastq.gz'  ftype="fastq.gz" />
            <param name='barcodeFile' value='scHicDemultiplex/GSE94489_README.txt' />
            <param name='srrToSample' value='scHicDemultiplex/samples.txt' />
            <param name='bufferSize' value='1000' />
            <output_collection name="outputFastqsForward" type="list" count="24">
                <!-- <element name="Diploid_15_AGGCAGAA_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_AGGCAGAA_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/> -->
                <element name="Diploid_15_AGGCAGAA_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_AGGCAGAA_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_CGTACTAG_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_CGTACTAG_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <!-- <element name="Diploid_15_AGGCAGAA_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_AGGCAGAA_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/> -->
                <element name="Diploid_15_GGACTCCT_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_GGACTCCT_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_TAAGGCGA_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_TAAGGCGA_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_TAGGCATG_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_TAGGCATG_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_TCCTGAGC_CTCTCTAT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_TCCTGAGC_CTCTCTAT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_AGGCAGAA_TATCCTCT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_AGGCAGAA_TATCCTCT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_CGTACTAG_TATCCTCT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_CGTACTAG_TATCCTCT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_GGACTCCT_TATCCTCT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_GGACTCCT_TATCCTCT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_TAAGGCGA_TATCCTCT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_TAAGGCGA_TATCCTCT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_TAGGCATG_TATCCTCT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_TAGGCATG_TATCCTCT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_TCCTGAGC_TATCCTCT_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_TCCTGAGC_TATCCTCT_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_AGGCAGAA_GTAAGGAG_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_AGGCAGAA_GTAAGGAG_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_CGTACTAG_GTAAGGAG_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_CGTACTAG_GTAAGGAG_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_GGACTCCT_GTAAGGAG_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_GGACTCCT_GTAAGGAG_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_TAAGGCGA_GTAAGGAG_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_TAAGGCGA_GTAAGGAG_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_TAGGCATG_GTAAGGAG_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_TAGGCATG_GTAAGGAG_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_TCCTGAGC_GTAAGGAG_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_TCCTGAGC_GTAAGGAG_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_AGGCAGAA_ACTGCATA_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_AGGCAGAA_ACTGCATA_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_CGTACTAG_ACTGCATA_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_CGTACTAG_ACTGCATA_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_GGACTCCT_ACTGCATA_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_GGACTCCT_ACTGCATA_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_TAAGGCGA_ACTGCATA_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_TAAGGCGA_ACTGCATA_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_TAGGCATG_ACTGCATA_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_TAGGCATG_ACTGCATA_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_TCCTGAGC_ACTGCATA_R1.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_TCCTGAGC_ACTGCATA_R1.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
            </output_collection>

            <output_collection name="outputFastqsReverse" type="list" count="24">
                <element name="Diploid_15_AGGCAGAA_CTCTCTAT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_AGGCAGAA_CTCTCTAT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_CGTACTAG_CTCTCTAT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_CGTACTAG_CTCTCTAT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_GGACTCCT_CTCTCTAT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_GGACTCCT_CTCTCTAT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_TAAGGCGA_CTCTCTAT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_TAAGGCGA_CTCTCTAT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_TAGGCATG_CTCTCTAT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_TAGGCATG_CTCTCTAT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_15_TCCTGAGC_CTCTCTAT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_15_TCCTGAGC_CTCTCTAT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_AGGCAGAA_TATCCTCT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_AGGCAGAA_TATCCTCT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_CGTACTAG_TATCCTCT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_CGTACTAG_TATCCTCT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_GGACTCCT_TATCCTCT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_GGACTCCT_TATCCTCT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_TAAGGCGA_TATCCTCT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_TAAGGCGA_TATCCTCT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_TAGGCATG_TATCCTCT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_TAGGCATG_TATCCTCT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_16_TCCTGAGC_TATCCTCT_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_16_TCCTGAGC_TATCCTCT_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_AGGCAGAA_GTAAGGAG_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_AGGCAGAA_GTAAGGAG_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_CGTACTAG_GTAAGGAG_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_CGTACTAG_GTAAGGAG_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_GGACTCCT_GTAAGGAG_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_GGACTCCT_GTAAGGAG_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_TAAGGCGA_GTAAGGAG_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_TAAGGCGA_GTAAGGAG_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_TAGGCATG_GTAAGGAG_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_TAGGCATG_GTAAGGAG_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_17_TCCTGAGC_GTAAGGAG_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_17_TCCTGAGC_GTAAGGAG_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_AGGCAGAA_ACTGCATA_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_AGGCAGAA_ACTGCATA_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_CGTACTAG_ACTGCATA_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_CGTACTAG_ACTGCATA_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_GGACTCCT_ACTGCATA_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_GGACTCCT_ACTGCATA_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_TAAGGCGA_ACTGCATA_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_TAAGGCGA_ACTGCATA_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_TAGGCATG_ACTGCATA_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_TAGGCATG_ACTGCATA_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
                <element name="Diploid_18_TCCTGAGC_ACTGCATA_R2.fastq.gz" file="scHicDemultiplex/demultiplexed/Diploid_18_TCCTGAGC_ACTGCATA_R2.fastq.gz"  ftype="fastqsanger.gz" compare='sim_size' delta='40000'/>
            </output_collection>
        </test>
    </tests>
    <help><![CDATA[

Demultiplexing of raw fastq files
=================================

scHicDemultiplex demultiplexes fastq files from Nagano 2017: "Cell-cycle dynamics of chromosomal organization at single-cell resolution" according their barcodes to
a seperated forward and reverse strand fastq files per cell. For other datasets, a third-party demultiplexing strategy must be used.

Afterwards, the demultiplexed mapped data can be used with HiCExplorer `hicBuildMatrix` to create single cell .cool matrices that must be stored in a .scool file using `scHicMergeToSCool`, in order to be used for downstream analyses in the scHiCExplorer suite.

For more information about scHiCExplorer please consider our documentation on readthedocs.io_

.. _readthedocs.io: http://schicexplorer.readthedocs.io/
]]></help>
    <expand macro="citations" />

</tool>