<tool id="sniplay_plink" name="PLINK" version="2.0.0">
    
    <!-- [REQUIRED] Tool description displayed after the tool name -->
    <description>Filter large VCF file</description>
    
    <!-- [OPTIONAL] 3rd party tools, binaries, modules... required for the tool to work -->
    <requirements>
        <requirement type="binary">perl</requirement>
        <requirement type="package" version="1.6.924">perl-bioperl</requirement>
	<requirement type="package" version="1.90b4">plink</requirement>
    </requirements>
    
    <code file="find_indiv.py"/>

    <!-- [STRONGLY RECOMMANDED] Exit code rules -->
    <stdio>
        <!-- [HELP] If no exit code rule is defined, the tool will stop if anything is written to STDERR -->
        <exit_code range="1:" level="fatal" />
    </stdio>    

    <!-- [REQUIRED] The command to execute -->
    <command interpreter="bash">
	./plink.sh $vcf $fileout $filelog $frequency $max_freq $allow_missing $type_p $bound_start $bound_end
	#if str( $samples ) == "":
	'None'
	#else
	$samples
	#end if
	#if str( $chromosomes ) == "":
	'None'
	#else
	$chromosomes
	#end if
    </command>
    <!-- [REQUIRED] Input files and tool parameters -->
    <inputs>
	<param name="vcf" type="data" format="vcf" optional="false" label="VCF input" />

	<param name="samples" type="select" label="Samples" multiple="true" dynamic_options="get_field_samples_options(vcf)" help="Samples to be analyzed." />
	<!--<param name="samples" type="text" label="Samples" multiple="true" help="Samples to be analyzed." />-->
        <!--<param name="chromosomes" type="select" label="Chromosomes" multiple="true" dynamic_options="get_field_chrs_options(input)" help="Chromosomes to be analyzed." />-->
	<param name="frequency" type="float" value="0" label="Minimum MAF" help="Minimum Minor Allele Frequency (MAF)" />
	<param name="max_freq" type="float" value="0.5" label="Maximum MAF" help="Maximum Minor Allele Frequency (MAF)" />
	<param name="allow_missing" type="float" value="1" min="0" max="1" label="Missing data proportion" help="Allowed missing data proportion per site. Must be comprised between 0 and 1." />
        <param name="type_p" type="select" label="Polymorphisms" help="Type of polymorphisms to keep." >
            <option value="ALL" selected="true">All</option>
            <option value="SNP">SNPs only</option>
        </param>
        <param name="chromosomes" type="text" label="Chromosomes" multiple="true" help="Chromosomes to be analyzed. (Comma-separated list of reference sequences, ex: Chr1,Chr2)" />
	<param name="bound_start" type="integer" value="1" label="Lower bound" help="Lower bound for a range of sites to be processed." />
	<param name="bound_end" type="integer" value="100000000" label="Upper bound" help="Upper bound for a range of sites to be processed." />
    </inputs>
    
    <!-- [REQUIRED] Output files -->
    <outputs>
	<data name="fileout" format="vcf" label="PLINK filtered VCF"/>
	<data name="filelog" format="txt" label="PLINK logfile" />
    </outputs>
    
    <tests>
        <test>
            <param name="vcf" value="plink_input.vcf"/>
            <param name="samples" value="AZUCENA,BULUPANDAK,GIZA171,IAC165,KHAODAM,M202,MOROBEREKAN,NIPPONBARE.D,62667,63-104,ARAGUAIA,ARIAS,ARROZCEBADA,BABER,BAGANANASALAO,BAKUNGH,BENGALYVAKARINA,BICOBRANCO,BINULAWAN,CAIAPO,CANAROXA,CANELADEFERRO,CHALOYOE,CHAPHUMA,CICIHBETON,CIRAD358,CIRAD392,CIRAD394,CIRAD403,CIRAD409,CIRAD488,CNA-7.BO.1.1.33-13-6-1,COLOMBIA1,CT13582-15-5-M,CUBA65,CUIABANA,CURINCA,DAM,DANGREY,DAVAO,DAWASANRED,DINORADO,DOURADOAGULHA,DOURADOPRECOCE,ESPERANZA,FOHISOMOTRA,GANIGI,GEMJYAJYANAM,GOGO,GOGOLEMPUK,GOGOLEMPAK,GOMPA2,GRAZI,GUARANI,GUNDILKUNING,HAWMOM,HD1-4,IAC25,IAC47,IDSA77,IGUAPECATETO,INDANE,IR60080-46A,IR63380-16,IR63372-08,IR65907-188-1-B,IR66421-096-2-1-1,IR68704-145-1-1-B,IR71525-19-1-1,IRAT104,IRAT109,IRAT112,IRAT13,IRAT144,IRAT170,IRAT177,IRAT2,IRAT212,IRAT216,IRAT234,IRAT257,IRAT335,IRAT362,IRAT364,IRAT366,IRAT380,JAOHAW,JIMBRUKJOLOWORO,JUMALI,JUMULA2,KAKANI2,KANIRANGA,KARASUKARASURANKASU,KEDAYAN,KENDINGA5H,KETANKONIR,KETANLUMBU,KETANMENAH,KHAOKAPXANG,KINANDANGPATONG,KOMOJAMANITRA,KU115,KUROKA,LAMBAYQUE1,LUDAN,MAHAE,MAINTIMOLOTSY1226,MALAGKITPIRURUTONG,MANANELATRA520,MANDRIRAVINA3512,MARAVILHA,MITSANGANAHIJERY,MOLOK,NABESHI,NEPHOAVANG,NHTA10,NHTA5,NPE253,NPE826,ORYZICASABANA6,OS4,OS6,P5589-1-1-3-P,PACHOLINHA,PADIBOENAR,PADIKASALLE,PALAWAN,PATEBLANCMAN1,PCT11.0.0.2.BO.1.55-1-3-1,PCT4.SA.4.1.1076-2-4-1-5,PEHPINUO,POENOETHITAM,PULULAPA,RATHAL,REKETMAUN,RT1031-69,SENG,SPEAKER,TANDUI,TREMBESE,TRESMESES,TSIPALA89,VARYLAVA90,VARYLAVADEBETAFO,VARYMADINIKA3566,VARYMALADY,VARYSOMOTRASIHANAKA,WAB56-125,WAB56-50,WAB706-3-4-K4-KB-1,YANCAOUSSA,YANGKUMRED,YUNLU7,IR47686-09-01-B-1,IR53236-275-1,IR65261-19-1-B,IR65907-206-4-B,IR66421-105-1-1,IR71524-44-1-1,VIETNAM1,VIETNAM3,IR47684-05-1-B,PRIMAVERA,CHUAN3,CHUAN4,EARLYMUTANTIAC165"/>
            <param name="frequency" value="0"/>
            <param name="max_freq" value="0.5"/>
            <param name="allow_missing" value="1"/>
            <param name="type_p" value="ALL"/>
            <param name="chromosomes" value=""/>
            <param name="bound_start" value="1"/>
            <param name="bound_end" value="100000000"/>
            <output name="fileout" file="plink_output.vcf" compare="diff" lines_diff="2"/>
        </test>
    </tests> 
    
    <!-- [OPTIONAL] Help displayed in Galaxy -->
    <help><![CDATA[

.. class:: infomark

**Authors**  Shaun Purcell : https://www.cog-genomics.org/plink2

 | **Please cite** "PLINK: a toolset for whole-genome association and population-based linkage analysis.", Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MAR, Bender D, Maller J, Sklar P, de Bakker PIW, Daly MJ, Sham PC, **American Journal of Human Genetics**, 2007 

.. class:: infomark

**Galaxy integration** Provided by Southgreen & Dereeper Alexis (IRD) & Marcon Valentin (IFB & INRA)

.. class:: infomark

**Support** For any questions about Galaxy integration, please send an e-mail to alexis.dereeper@ird.fr

---------------------------------------------------

=====
PLINK
=====

-----------
Description
-----------

  | Genome association analysis toolset, designed to perform a range of basic, large-scale analyses in a computationally efficient manner.  
  | For further informations, please visit the plink website_.

.. _website: https://www.cog-genomics.org/plink2

------------
Dependencies
------------

PLINK
        plink_ 1.90b4, Conda version
Bioperl
        perl-bioperl_ 1.6.924, Conda version

.. _plink: https://anaconda.org/bioconda/plink
.. _perl-bioperl: https://anaconda.org/bioconda/perl-bioperl

    ]]></help>
    <citations>
       <citation type="doi">10.1086/519795</citation>
       <citation type="doi">10.1186/s13742-015-0047-8</citation>
    </citations>
</tool>