kkonganti@92: process SEQKIT_GREP {
kkonganti@92:     tag "$meta.id"
kkonganti@92:     label 'process_low'
kkonganti@92: 
kkonganti@92:     module (params.enable_module ? "${params.swmodulepath}${params.fs}seqkit${params.fs}2.2.0" : null)
kkonganti@92:     conda (params.enable_conda ? "bioconda::seqkit=2.2.0 conda-forge::sed=4.7 conda-forge::coreutils" : null)
kkonganti@92:     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
kkonganti@92:         'https://depot.galaxyproject.org/singularity/seqkit:2.1.0--h9ee0642_0':
kkonganti@92:         'quay.io/biocontainers/seqkit:2.1.0--h9ee0642_0' }"
kkonganti@92: 
kkonganti@92:     input:
kkonganti@92:     tuple val(meta), path(reads), path(pattern_file)
kkonganti@92: 
kkonganti@92:     output:
kkonganti@92:     tuple val(meta), path("*.gz"), emit: fastx
kkonganti@92:     path "versions.yml"          , emit: versions
kkonganti@92: 
kkonganti@92:     when:
kkonganti@92:     task.ext.when == null || task.ext.when
kkonganti@92: 
kkonganti@92:     script:
kkonganti@92:     def args = task.ext.args ?: ''
kkonganti@92:     def prefix = task.ext.prefix ?: "${meta.id}"
kkonganti@92: 
kkonganti@92:     def extension = "fastq"
kkonganti@92:     if ("$reads" ==~ /.+\.fasta|.+\.fasta.gz|.+\.fa|.+\.fa.gz|.+\.fas|.+\.fas.gz|.+\.fna|.+\.fna.gz/) {
kkonganti@92:         extension = "fasta"
kkonganti@92:     }
kkonganti@92: 
kkonganti@92:     if (meta.single_end) {
kkonganti@92:         """
kkonganti@92:         pattern_file_contents=\$(sed '1!d' $pattern_file)
kkonganti@92:         if [ "\$pattern_file_contents" != "DuMmY" ]; then
kkonganti@92:             additional_args="-f $pattern_file $args"
kkonganti@92:         else
kkonganti@92:             additional_args="$args"
kkonganti@92:         fi
kkonganti@92: 
kkonganti@92:         seqkit \\
kkonganti@92:             grep \\
kkonganti@92:             -j $task.cpus \\
kkonganti@92:             -o ${prefix}.seqkit-grep.${extension}.gz \\
kkonganti@92:             \$additional_args \\
kkonganti@92:             $reads
kkonganti@92: 
kkonganti@92:         cat <<-END_VERSIONS > versions.yml
kkonganti@92:         "${task.process}":
kkonganti@92:             seqkit: \$( seqkit | sed '3!d; s/Version: //' )
kkonganti@92:         END_VERSIONS
kkonganti@92:         """
kkonganti@92:     } else {
kkonganti@92:         """
kkonganti@92:         pattern_file_contents=\$(sed '1!d' $pattern_file)
kkonganti@92:         if [ "\$pattern_file_contents" != "DuMmY" ]; then
kkonganti@92:             additional_args="-f $pattern_file $args"
kkonganti@92:         else
kkonganti@92:             additional_args="$args"
kkonganti@92:         fi
kkonganti@92: 
kkonganti@92:         seqkit \\
kkonganti@92:             grep \\
kkonganti@92:             -j $task.cpus \\
kkonganti@92:             -o ${prefix}.R1.seqkit-grep.${extension}.gz \\
kkonganti@92:             \$additional_args \\
kkonganti@92:             ${reads[0]}
kkonganti@92:         
kkonganti@92:         seqkit \\
kkonganti@92:             grep \\
kkonganti@92:             -j $task.cpus \\
kkonganti@92:             -o ${prefix}.R2.seqkit-grep.${extension}.gz \\
kkonganti@92:             \$additional_args \\
kkonganti@92:             ${reads[1]}
kkonganti@92: 
kkonganti@92:         seqkit \\
kkonganti@92:             pair \\
kkonganti@92:             -j $task.cpus \\
kkonganti@92:             -1 ${prefix}.R1.seqkit-grep.${extension}.gz \\
kkonganti@92:             -2 ${prefix}.R2.seqkit-grep.${extension}.gz
kkonganti@92: 
kkonganti@92:         rm ${prefix}.R1.seqkit-grep.${extension}.gz
kkonganti@92:         rm ${prefix}.R2.seqkit-grep.${extension}.gz
kkonganti@92: 
kkonganti@92:         cat <<-END_VERSIONS > versions.yml
kkonganti@92:         "${task.process}":
kkonganti@92:             seqkit: \$( seqkit | sed '3!d; s/Version: //' )
kkonganti@92:         END_VERSIONS
kkonganti@92:         """
kkonganti@92:     }
kkonganti@92: }