view qiime2/qiime_tools_import.xml @ 0:370e0b6e9826 draft

Uploaded
author florianbegusch
date Wed, 17 Jul 2019 03:05:17 -0400
parents
children 914fa4daf16a
line wrap: on
line source

<?xml version="1.0" ?>
<tool id="qiime_tools_import" name="qiime tools import" version="2019.4">
	<description> - Import data into a new QIIME 2 Artifact.</description>
	<requirements>
		<requirement type="package" version="2019.4">qiime2</requirement>
	</requirements>
	<command>
<![CDATA[
mkdir input;

#set $in_= 'input'


#if '__ob__' in str($semantic_type):
	#set $type_temp = str($semantic_type).replace('__ob__', '[')
	#set $type_temp = str($type_temp).replace('__cb__', ']')
	#set $semantic_type = $type_temp
#end if

#def is_fastq($f_test)
	#set $f = open($f_test, 'rb')
	#if str($f.readline()[0]) == '@'
		#$f.close()
		#return True
	#end if

	#$f.close()
	#return False
#end def


## Symlinks creation or On the fly creation of a combo file
#if $input_type.type == 'list_list'
	## Symlinks creation or On the fly creation of a combo file
	#for $f in $input_type.list_list:

		#set $is_fastq_file = $is_fastq(str($f))

		#if $is_fastq_file:
			#set $file_for_processing = "'input/" + $f.name + '.gz' + "'"
			gzip -c ${f} > $file_for_processing;
		#else
			#set $file_for_processing = "'input/" + $f.name + "'"
			ln -s ${f}  $file_for_processing;
		#end if
	#end for
	#if 'SingleEndFastqManifestPhred' in str($sourceformat):
			#set $cwd = str($outputpath.extra_files_path) + "_manifest_file.txt"
			#set $m_file = open(str($cwd), 'w')
			$m_file.write("sample-id,absolute-filepath,direction\n")
			#for $f in $input_type.list_list:
				$m_file.write(str($f.name).split("_")[0]+",$PWD/input/"+str($f.name)+".gz,forward\n")
			#end for
			$m_file.close()
			#set $in_= str($cwd)
		#end if
	

#elif $input_type.type == 'list_paired_collection':
  
	#set $cwd = str($outputpath.extra_files_path) + "_manifest_file.txt"
	#set $m_file = open(str($cwd), 'w')
	$m_file.write("sample-id,absolute-filepath,direction\n")	

	#for $pair in $input_type.list_paired_collection:
		#set $forward_is_fastq_file = $is_fastq(str($pair.forward))
		#set $reverse_is_fastq_file = $is_fastq(str($pair.reverse))
		
		#if $forward_is_fastq_file:
	 		gzip -c ${pair.forward} > input/${pair.name}_R1_001.fastq.gz;
	 		$m_file.write(str($pair.name).split("_")[0]+",$PWD/input/"+str($pair.name)+"_R1_001.fastq.gz,forward\n")
		#else
	  		ln -s ${pair.forward} input/${pair.name}_R1_001.fastq.gz;
			$m_file.write(str($pair.name).split("_")[0]+",$PWD/input/"+str($pair.name)+"_R1_001.fastq.gz,forward\n")
        #end if

	 	#if $reverse_is_fastq_file:
	 		gzip -c ${pair.reverse} > input/${pair.name}_R2_001.fastq.gz;
			$m_file.write(str($pair.name).split("_")[0]+",$PWD/input/"+str($pair.name)+"_R2_001.fastq.gz,reverse\n")
        #else
	  		ln -s ${pair.reverse} input/${pair.name}_R2_001.fastq.gz;
			$m_file.write(str($pair.name).split("_")[0]+",$PWD/input/"+str($pair.name)+"_R2_001.fastq.gz,reverse\n")
	  	#end if
	#end for

	
	$m_file.close()
	#if 'PairedEndFastqManifestPhred' in str($sourceformat):
		#set $in_= str($cwd)
	#elif  'Casava' in str($sourceformat):
		#set $in_= 'input'
	#end if

#else
	#set $res = $is_fastq(str($file))

	#if $res
		gzip -c ${file} > ${file.name}.gz;
		#set $in_= $file.name + '.gz'

	#else
		#set $in_= str($file)
	#end if
   
   #if "MultiplexedSingleEndBarcodeInSequence" in str($semantic_type):      
 			mv *_R1_* forward.fastq.gz;
	    #set $in_= 'forward.fastq.gz';
   #end if 

#end if


## rename files to predefined names for "MultiplexedPairedEndBarcodeInSequence"
#if "MultiplexedPairedEndBarcodeInSequence" in str($semantic_type):
		mv input/*_R1_* input/forward.fastq.gz;
		mv input/*_R2_* input/reverse.fastq.gz;
#end if


qiime tools import

--type="$semantic_type"

--input-path=$in_

--output-path=outputpath.qza

#if str($sourceformat) != 'None':
	#if '__ob__' in str($sourceformat):
  		#set $sourceformat_temp = str($sourceformat).replace('__ob__', '[')
		#set $sourceformat_temp = str($sourceformat_temp).replace('__cb__', ']')
  		#set $sourceformat = $sourceformat_temp
	#end if
 --source-format="$sourceformat"
#end if
;
cp outputpath.qza $outputpath
]]>
	</command>
	<inputs>
		<conditional name="input_type">
			<param name="type" type="select" label="Select a way to provide data, either using a file or a list" help="">
				<option value="regular">Input file</option>
                <option selected="true" value="list_paired_collection">Multiple pairs - Combo mode (list:paired collection)</option>
				<option value="list_list">Input list</option>
			</param>

			<when value="regular">
				<param name="file" type="data" multiple="false" label="Input file" />
			</when>
		 	<when value="list_paired_collection">
            	<param name="list_paired_collection" type="data_collection" collection_type="list:paired" label="Fastq pairs (collection)" help="Dataset collection made from multiple pairs of files" />
         	</when>
			<when value="list_list">
				<param name="list_list" type="data_collection" collection_type="list" label="Input list" />
			</when>
		</conditional>

		<param label="--type: The semantic type of the artifact that will be created upon importing. [required]" name="semantic_type" optional="False" type="select">
			<option value="DeblurStats">DeblurStats</option>
			<option value="DistanceMatrix">DistanceMatrix</option>
			<option value="EMPPairedEndSequences">EMPPairedEndSequences</option>
			<option value="EMPSingleEndSequences">EMPSingleEndSequences</option>
			<option value="FeatureData[AlignedSequence]">FeatureData[AlignedSequence]</option>
			<option value="FeatureData[PairedEndSequence]">FeatureData[PairedEndSequence]</option>
			<option value="FeatureData[Sequence]">FeatureData[Sequence]</option>
			<option value="FeatureData[Taxonomy]">FeatureData[Taxonomy]</option>
			<option value="FeatureTable[Balance]">FeatureTable[Balance]</option>
			<option value="FeatureTable[Composition]">FeatureTable[Composition]</option>
			<option value="FeatureTable[Frequency]">FeatureTable[Frequency]</option>
			<option value="FeatureTable[PresenceAbsence]">FeatureTable[PresenceAbsence]</option>
			<option value="FeatureTable[RelativeFrequency]">FeatureTable[RelativeFrequency]</option>
			<option value="Hierarchy">Hierarchy</option>
			<option value="MultiplexedPairedEndBarcodeInSequence">MultiplexedPairedEndBarcodeInSequence</option>
			<option value="MultiplexedSingleEndBarcodeInSequence">MultiplexedSingleEndBarcodeInSequence</option>
			<option value="PCoAResults">PCoAResults</option>
			<option value="Phylogeny[Rooted]">Phylogeny[Rooted]</option>
			<option value="Phylogeny[Unrooted]">Phylogeny[Unrooted]</option>
			<option value="QualityFilterStats">QualityFilterStats</option>
			<option value="RawSequences">RawSequences</option>
			<option value="SampleData[AlphaDiversity]">SampleData[AlphaDiversity]</option>
			<option value="SampleData[BooleanSeries]">SampleData[BooleanSeries]</option>
			<option value="SampleData[DADA2Stats]">SampleData[DADA2Stats]</option>
			<option value="SampleData[FirstDifferences]">SampleData[FirstDifferences]</option>
			<option value="SampleData[JoinedSequencesWithQuality]">SampleData[JoinedSequencesWithQuality]</option>
			<option selected="True" value="SampleData[PairedEndSequencesWithQuality]">SampleData[PairedEndSequencesWithQuality]</option>
			<option value="SampleData[SequencesWithQuality]">SampleData[SequencesWithQuality]</option>
			<option value="SampleData[Sequences]">SampleData[Sequences]</option>
			<option value="TaxonomicClassifier">TaxonomicClassifier</option>
			<option value="UchimeStats">UchimeStats</option>
		</param>

		<param label="--source-format: The format of the data to be imported. If not provided, data must be in the format expected by the semantic type provided via --type." name="sourceformat" optional="True" type="select">
			<option value="AlignedDNAFASTAFormat">AlignedDNAFASTAFormat</option>
			<option value="AlignedDNASequencesDirectoryFormat">AlignedDNASequencesDirectoryFormat</option>
			<option value="AlphaDiversityDirectoryFormat">AlphaDiversityDirectoryFormat</option>
			<option value="AlphaDiversityFormat">AlphaDiversityFormat</option>
			<option value="BIOMV100DirFmt">BIOMV100DirFmt</option>
			<option value="BIOMV100Format">BIOMV100Format</option>
			<option value="BIOMV210DirFmt">BIOMV210DirFmt</option>
			<option value="BIOMV210Format">BIOMV210Format</option>
			<option value="BooleanSeriesDirectoryFormat">BooleanSeriesDirectoryFormat</option>
			<option value="BooleanSeriesFormat">BooleanSeriesFormat</option>
			<option value="CasavaOneEightLanelessPerSampleDirFmt">CasavaOneEightLanelessPerSampleDirFmt</option>
			<option value="CasavaOneEightSingleLanePerSampleDirFmt">CasavaOneEightSingleLanePerSampleDirFmt</option>
			<option value="DADA2StatsDirFmt">DADA2StatsDirFmt</option>
			<option value="DADA2StatsFormat">DADA2StatsFormat</option>
			<option value="DNAFASTAFormat">DNAFASTAFormat</option>
			<option value="DNASequencesDirectoryFormat">DNASequencesDirectoryFormat</option>
			<option value="DeblurStatsDirFmt">DeblurStatsDirFmt</option>
			<option value="DeblurStatsFmt">DeblurStatsFmt</option>
			<option value="DistanceMatrixDirectoryFormat">DistanceMatrixDirectoryFormat</option>
			<option value="EMPPairedEndCasavaDirFmt">EMPPairedEndCasavaDirFmt</option>
			<option value="EMPPairedEndDirFmt">EMPPairedEndDirFmt</option>
			<option value="EMPSingleEndCasavaDirFmt">EMPSingleEndCasavaDirFmt</option>
			<option value="EMPSingleEndDirFmt">EMPSingleEndDirFmt</option>
			<option value="FastqGzFormat">FastqGzFormat</option>
			<option value="FirstDifferencesDirectoryFormat">FirstDifferencesDirectoryFormat</option>
			<option value="FirstDifferencesFormat">FirstDifferencesFormat</option>
			<option value="HeaderlessTSVTaxonomyDirectoryFormat">HeaderlessTSVTaxonomyDirectoryFormat</option>
			<option value="HeaderlessTSVTaxonomyFormat">HeaderlessTSVTaxonomyFormat</option>
			<option value="LSMatFormat">LSMatFormat</option>
			<option value="MultiplexedPairedEndBarcodeInSequenceDirFmt">MultiplexedPairedEndBarcodeInSequenceDirFmt</option>
			<option value="MultiplexedSingleEndBarcodeInSequenceDirFmt">MultiplexedSingleEndBarcodeInSequenceDirFmt</option>
			<option value="NewickDirectoryFormat">NewickDirectoryFormat</option>
			<option value="NewickFormat">NewickFormat</option>
			<option value="OrdinationDirectoryFormat">OrdinationDirectoryFormat</option>
			<option value="OrdinationFormat">OrdinationFormat</option>
			<option value="PairedDNASequencesDirectoryFormat">PairedDNASequencesDirectoryFormat</option>
			<option selected="True" value="PairedEndFastqManifestPhred33">PairedEndFastqManifestPhred33</option>
			<option value="PairedEndFastqManifestPhred64">PairedEndFastqManifestPhred64</option>
			<option value="QIIME1DemuxDirFmt">QIIME1DemuxDirFmt</option>
			<option value="QIIME1DemuxFormat">QIIME1DemuxFormat</option>
			<option value="QualityFilterStatsDirFmt">QualityFilterStatsDirFmt</option>
			<option value="QualityFilterStatsFmt">QualityFilterStatsFmt</option>
			<option value="SingleEndFastqManifestPhred33">SingleEndFastqManifestPhred33</option>
			<option value="SingleEndFastqManifestPhred64">SingleEndFastqManifestPhred64</option>
			<option value="SingleLanePerSamplePairedEndFastqDirFmt">SingleLanePerSamplePairedEndFastqDirFmt</option>
			<option value="SingleLanePerSampleSingleEndFastqDirFmt">SingleLanePerSampleSingleEndFastqDirFmt</option>
			<option value="TSVTaxonomyDirectoryFormat">TSVTaxonomyDirectoryFormat</option>
			<option value="TSVTaxonomyFormat">TSVTaxonomyFormat</option>
			<option value="TaxonomicClassiferTemporaryPickleDirFmt">TaxonomicClassiferTemporaryPickleDirFmt</option>
			<option value="UchimeStatsDirFmt">UchimeStatsDirFmt</option>
			<option value="UchimeStatsFmt">UchimeStatsFmt</option>
		</param>
	</inputs>
	<outputs>
		<data format="qza" label="${tool.name} on ${on_string}: outputpath.qza" name="outputpath"/>
	</outputs>
	    <stdio>
        <exit_code range="2"   level="warning"   description="Error" />
    </stdio>
	<help>
</help>
<macros>
	<import>qiime_citation.xml</import>
</macros>
<expand macro="qiime_citation" />
</tool>