1
0
Fork 0
mirror of https://github.com/MillironX/taxprofiler.git synced 2024-09-21 14:42:04 +00:00
taxprofiler/conf/modules.config

556 lines
20 KiB
Text
Raw Normal View History

/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Config file for defining DSL2 per module options and publishing paths
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Available keys to override module options:
ext.args = Additional arguments appended to command in module.
ext.args2 = Second set of arguments appended to command in module (multi-tool modules).
ext.args3 = Third set of arguments appended to command in module (multi-tool modules).
ext.prefix = File name prefix for output files.
----------------------------------------------------------------------------------------
*/
process {
withName: FASTQC {
ext.args = '--quiet'
ext.prefix = { "${meta.id}_${meta.run_accession}_raw" }
publishDir = [
path: { "${params.outdir}/fastqc/raw" },
mode: params.publish_dir_mode,
pattern: '*.html'
]
}
2022-03-25 13:58:06 +00:00
withName: FASTQC_PROCESSED {
ext.args = '--quiet'
ext.prefix = { "${meta.id}_${meta.run_accession}_processed" }
publishDir = [
path: { "${params.outdir}/fastqc/processed" },
mode: params.publish_dir_mode,
2022-03-25 13:58:06 +00:00
pattern: '*.html'
]
}
2022-10-18 15:43:16 +00:00
withName: FALCO {
2022-12-15 11:57:49 +00:00
ext.prefix = { "${meta.id}_${meta.run_accession}_raw_falco" }
2022-10-18 15:43:16 +00:00
publishDir = [
path: { "${params.outdir}/falco/raw" },
mode: params.publish_dir_mode,
pattern: '*.{html,txt}'
]
}
withName: FALCO_PROCESSED {
2022-12-15 11:57:49 +00:00
ext.prefix = { "${meta.id}_${meta.run_accession}_processed_falco" }
2022-10-18 15:43:16 +00:00
publishDir = [
path: { "${params.outdir}/falco/processed" },
mode: params.publish_dir_mode,
pattern: '*.{html,txt}'
]
}
withName: FASTP_SINGLE {
ext.args = [
// trimming options
params.shortread_qc_skipadaptertrim ? "--disable_adapter_trimming" : "",
2022-11-02 11:39:34 +00:00
params.shortread_qc_adapterlist ? "" : params.shortread_qc_adapter1 ? "--adapter_sequence ${params.shortread_qc_adapter1}" : "",
// filtering options
2022-05-31 18:11:19 +00:00
"--length_required ${params.shortread_qc_minlength}",
(params.perform_shortread_complexityfilter && params.shortread_complexityfilter_tool == 'fastp') ? "--low_complexity_filter --complexity_threshold ${params.shortread_complexityfilter_fastp_threshold}" : ''
].join(' ').trim()
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/fastp" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_preprocessed_reads
],
[
path: { "${params.outdir}/fastp" },
mode: params.publish_dir_mode,
pattern: '*.{log,html,json}'
]
]
}
withName: FASTP_PAIRED {
ext.args = [
2022-03-27 07:30:23 +00:00
// collapsing options - option to retain singletons
params.shortread_qc_includeunmerged ? '--include_unmerged' : '',
2022-03-25 13:58:06 +00:00
// trimming options
params.shortread_qc_skipadaptertrim ? "--disable_adapter_trimming" : "",
2022-11-02 11:39:34 +00:00
params.shortread_qc_adapterlist ? "" : params.shortread_qc_adapter1 ? "--adapter_sequence ${params.shortread_qc_adapter1}" : "",
params.shortread_qc_adapterlist ? "" : params.shortread_qc_adapter2 ? "--adapter_sequence_r2 ${params.shortread_qc_adapter2}" : "--detect_adapter_for_pe",
2022-03-25 13:58:06 +00:00
// filtering options
2022-05-31 18:11:19 +00:00
"--length_required ${params.shortread_qc_minlength}",
2022-05-07 04:12:24 +00:00
params.perform_shortread_complexityfilter && params.shortread_complexityfilter_tool == 'fastp' ? "--low_complexity_filter --complexity_threshold ${params.shortread_complexityfilter_fastp_threshold}" : ''
].join(' ').trim()
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/fastp" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_preprocessed_reads
],
[
path: { "${params.outdir}/fastp" },
mode: params.publish_dir_mode,
pattern: '*.{log,html,json}'
]
]
}
withName: ADAPTERREMOVAL_SINGLE {
ext.args = [
// trimming options
params.shortread_qc_skipadaptertrim ? "--adapter1 ''" : params.shortread_qc_adapterlist ? "" : params.shortread_qc_adapter1 ? "--adapter1 ${params.shortread_qc_adapter1}" : "",
// filtering options
"--minlength ${params.shortread_qc_minlength}"
].join(' ').trim()
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/adapterremoval" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_preprocessed_reads
],
[
path: { "${params.outdir}/adapterremoval" },
mode: params.publish_dir_mode,
pattern: '*.settings'
]
]
}
withName: ADAPTERREMOVAL_PAIRED {
ext.args = [
// collapsing options
params.shortread_qc_mergepairs ? "--collapse" : "",
// trimming options
params.shortread_qc_skipadaptertrim ? "--adapter1 ''" : params.shortread_qc_adapterlist ? "" : params.shortread_qc_adapter1 ? "--adapter1 ${params.shortread_qc_adapter1}" : "", // adding adapter list happens at module input channel level
params.shortread_qc_skipadaptertrim ? "--adapter2 ''" : params.shortread_qc_adapterlist ? "" : params.shortread_qc_adapter2 ? "--adapter2 ${params.shortread_qc_adapter2}" : "",
// filtering options
"--minlength ${params.shortread_qc_minlength}"
].join(' ').trim()
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/adapterremoval" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_preprocessed_reads
],
[
path: { "${params.outdir}/adapterremoval" },
mode: params.publish_dir_mode,
pattern: '*.settings'
]
]
}
withName: PORECHOP_PORECHOP {
2022-03-18 13:34:10 +00:00
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/porechop" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_preprocessed_reads
],
[
path: { "${params.outdir}/porechop" },
mode: params.publish_dir_mode,
pattern: '*.log'
]
2022-03-18 13:34:10 +00:00
]
}
withName: FILTLONG {
ext.args = [
"--min_length ${params.longread_qc_qualityfilter_minlength}",
"--keep_percent ${params.longread_qc_qualityfilter_keeppercent}",
"--target_bases ${params.longread_qc_qualityfilter_targetbases}"
]
.join(' ').trim()
2022-05-31 18:11:19 +00:00
ext.prefix = { "${meta.id}_${meta.run_accession}_filtered" }
publishDir = [
[
path: { "${params.outdir}/filtlong" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_preprocessed_reads
],
[
path: { "${params.outdir}/filtlong" },
mode: params.publish_dir_mode,
pattern: '*.log'
]
]
}
withName: BOWTIE2_BUILD {
publishDir = [
path: { "${params.outdir}/bowtie2/build" },
2022-04-07 12:00:10 +00:00
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_index,
2022-04-18 04:37:23 +00:00
pattern: 'bowtie2'
]
}
withName: BOWTIE2_ALIGN {
ext.prefix = { "${meta.id}_${meta.run_accession}" }
2022-04-18 04:37:23 +00:00
publishDir = [
[
path: { "${params.outdir}/bowtie2/align" },
mode: params.publish_dir_mode,
pattern: '*.log'
2022-04-18 04:37:23 +00:00
],
[
path: { "${params.outdir}/bowtie2/align" },
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_mapped,
pattern: '*.bam'
2022-04-18 04:37:23 +00:00
],
[
path: { "${params.outdir}/bowtie2/align" },
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_unmapped,
pattern: '*.fastq.gz'
2022-04-18 04:37:23 +00:00
]
]
}
2022-05-03 07:28:46 +00:00
withName: MINIMAP2_INDEX {
ext.args = '-x map-ont'
publishDir = [
path: { "${params.outdir}/minimap2/index" },
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_index,
2022-05-03 07:28:46 +00:00
pattern: 'minimap2'
]
}
withName: MINIMAP2_ALIGN {
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
path: { "${params.outdir}/minimap2/align" },
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_mapped,
2022-05-03 07:28:46 +00:00
pattern: '*.bam'
]
}
withName: SAMTOOLS_VIEW {
ext.args = '-f 4'
ext.prefix = { "${meta.id}.mapped.sorted" }
publishDir = [
path: { "${params.outdir}/samtools/view" },
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_unmapped,
2022-05-03 07:28:46 +00:00
pattern: '*.bam'
]
}
2022-10-31 11:35:51 +00:00
2022-05-03 07:28:46 +00:00
withName: SAMTOOLS_BAM2FQ {
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
path: { "${params.outdir}/samtools/bam2fq" },
mode: params.publish_dir_mode,
enabled: params.save_hostremoval_unmapped,
2022-05-03 07:28:46 +00:00
pattern: '*.fq.gz'
]
}
2022-10-31 11:17:49 +00:00
withName: SAMTOOLS_STATS {
2022-10-28 12:24:33 +00:00
ext.prefix = { "${meta.id}_${meta.run_accession}" }
publishDir = [
path: { "${params.outdir}/samtools/stats" },
mode: params.publish_dir_mode,
pattern: '*stats'
]
}
withName: BBMAP_BBDUK {
ext.args = [
"entropy=${params.shortread_complexityfilter_entropy}",
"entropywindow=${params.shortread_complexityfilter_bbduk_windowsize}",
params.shortread_complexityfilter_bbduk_mask ? "entropymask=t" : "entropymask=f"
].join(' ').trim()
ext.prefix = { "${meta.id}-${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/bbduk/" },
mode: params.publish_dir_mode,
pattern: '*.{fastq.gz,log}',
enabled: params.save_complexityfiltered_reads
],
[
path: { "${params.outdir}/bbduk/" },
mode: params.publish_dir_mode,
pattern: '*.log'
]
]
}
withName: PRINSEQPLUSPLUS {
ext.args = [
params.shortread_complexityfilter_prinseqplusplus_mode == 'dust' ? "-lc_dust=${params.shortread_complexityfilter_prinseqplusplus_dustscore}" : "-lc_entropy=${params.shortread_complexityfilter_entropy}",
2022-08-23 12:24:57 +00:00
"-trim_qual_left=0 -trim_qual_left=0 -trim_qual_window=0 -trim_qual_step=0",
"-VERBOSE 2"
].join(' ').trim()
ext.prefix = { "${meta.id}-${meta.run_accession}" }
publishDir = [
[
path: { "${params.outdir}/prinseqplusplus/" },
mode: params.publish_dir_mode,
pattern: '*{_good_out.fastq.gz,_good_out_R1.fastq.gz,_good_out_R2.fastq.gz}',
enabled: params.save_complexityfiltered_reads
],
[
path: { "${params.outdir}/prinseqplusplus/" },
mode: params.publish_dir_mode,
pattern: '*.log'
]
]
}
2022-04-08 11:28:37 +00:00
withName: CAT_FASTQ {
2022-04-12 07:25:27 +00:00
ext.prefix = { "${meta.id}" }
2022-04-08 11:28:37 +00:00
publishDir = [
path: { "${params.outdir}/run_merging/" },
mode: params.publish_dir_mode,
pattern: '*.fastq.gz',
enabled: params.save_runmerged_reads
]
}
withName: MALT_RUN {
2022-11-24 13:44:04 +00:00
ext.args = { "${meta.db_params} -m ${params.malt_mode}" }
// one run with multiple samples, so fix ID to just db name to ensure clean log name
ext.prefix = { "${meta.db_name}" }
publishDir = [
path: { "${params.outdir}/malt/${meta.db_name}/" },
mode: params.publish_dir_mode,
pattern: '*.{rma6,log,sam}'
]
}
2022-09-02 06:03:05 +00:00
withName: 'MEGAN_RMA2INFO_TSV' {
ext.args = "-c2c Taxonomy"
ext.prefix = { "${meta.id}" }
publishDir = [
path: { "${params.outdir}/malt/${meta.db_name}/" },
mode: params.publish_dir_mode,
pattern: '*.{txt.gz,megan}'
]
}
withName: KRAKEN2_KRAKEN2 {
ext.args = params.kraken2_save_minimizers ? { "${meta.db_params} --report-minimizer-data" } : { "${meta.db_params}" }
2022-12-15 12:25:01 +00:00
ext.prefix = params.perform_runmerging ? { meta.tool == "bracken" ? "${meta.id}-${meta.db_name}.bracken" : "${meta.id}-${meta.db_name}" } : { meta.tool == "bracken" ? "${meta.id}-${meta.run_accession}-${meta.db_name}.bracken" : "${meta.id}-${meta.run_accession}-${meta.db_name}" }
publishDir = [
path: { "${params.outdir}/kraken2/${meta.db_name}/" },
mode: params.publish_dir_mode,
pattern: '*.{txt,report,fastq.gz}'
]
}
withName: BRACKEN_BRACKEN {
errorStrategy = 'ignore'
2022-12-15 12:25:01 +00:00
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}.bracken" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}.bracken" }
publishDir = [
path: { "${params.outdir}/bracken/${meta.db_name}/" },
mode: params.publish_dir_mode,
pattern: '*.tsv'
]
}
withName: BRACKEN_COMBINEBRACKENOUTPUTS {
ext.prefix = { "bracken_${meta.id}_combined_reports" }
publishDir = [
path: { "${params.outdir}/bracken/" },
mode: params.publish_dir_mode,
pattern: '*.txt'
]
}
withName: KRAKENTOOLS_COMBINEKREPORTS_KRAKEN {
2022-09-13 14:47:19 +00:00
ext.prefix = { "kraken2_${meta.id}_combined_reports" }
publishDir = [
path: { "${params.outdir}/kraken2/" },
mode: params.publish_dir_mode,
pattern: '*.txt'
]
}
2022-10-25 09:50:44 +00:00
withName: KRAKENUNIQ_PRELOADEDKRAKENUNIQ {
ext.args = { "${meta.db_params}" }
// one run with multiple samples, so fix ID to just db name to ensure clean log name
ext.prefix = { "${meta.db_name}" }
2022-10-25 09:50:44 +00:00
publishDir = [
path: { "${params.outdir}/krakenuniq/${meta.db_name}/" },
mode: params.publish_dir_mode,
pattern: '*.{txt,report,fastq.gz}'
]
}
withName: KRONA_CLEANUP {
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
publishDir = [
path: { "${params.outdir}/krona/" },
mode: params.publish_dir_mode,
pattern: '*.{html}'
]
}
withName: KRONA_KTIMPORTTEXT {
ext.prefix = { "${meta.tool}-${meta.id}" }
publishDir = [
path: { "${params.outdir}/krona/" },
mode: params.publish_dir_mode,
pattern: '*.{html}'
]
}
2022-09-02 06:03:05 +00:00
withName: 'MEGAN_RMA2INFO_KRONA' {
ext.args = { "--read2class Taxonomy" }
ext.prefix = { "${meta.id}-${meta.db_name}" }
}
withName: KRONA_KTIMPORTTAXONOMY {
ext.args = "-i"
ext.prefix = { "${meta.tool}-${meta.id}" }
publishDir = [
path: { "${params.outdir}/krona/" },
mode: params.publish_dir_mode,
pattern: '*.{html}'
]
}
withName: METAPHLAN3_METAPHLAN3 {
2022-04-08 11:28:37 +00:00
ext.args = { "${meta.db_params}" }
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
2022-04-02 12:44:27 +00:00
publishDir = [
path: { "${params.outdir}/metaphlan3/${meta.db_name}/" },
2022-04-02 12:44:27 +00:00
mode: params.publish_dir_mode,
pattern: '*.{biom,txt}'
]
}
withName: METAPHLAN3_MERGEMETAPHLANTABLES {
2022-09-15 10:30:39 +00:00
ext.prefix = { "metaphlan3_${meta.id}_combined_reports" }
publishDir = [
path: { "${params.outdir}/metaphlan3/" },
mode: params.publish_dir_mode,
pattern: '*.{txt}'
2022-04-02 12:44:27 +00:00
]
}
withName: CENTRIFUGE_CENTRIFUGE {
publishDir = [
path: { "${params.outdir}/centrifuge/${meta.db_name}/" },
mode: params.publish_dir_mode,
pattern: '*.{txt,sam,gz}'
]
ext.args = { "${meta.db_params}" }
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}.centrifuge" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}.centrifuge" }
}
2022-04-22 13:24:10 +00:00
withName: CENTRIFUGE_KREPORT {
errorStrategy = {task.exitStatus == 255 ? 'ignore' : 'retry'}
2022-04-22 13:24:10 +00:00
ext.args = { "${meta.db_params}" }
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}.centrifuge" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}.centrifuge" }
2022-04-22 13:24:10 +00:00
publishDir = [
path: { "${params.outdir}/centrifuge/${meta.db_name}/" },
2022-04-22 13:24:10 +00:00
mode: params.publish_dir_mode,
pattern: '*.{txt}'
]
}
2022-09-13 14:47:19 +00:00
withName: KRAKENTOOLS_COMBINEKREPORTS_CENTRIFUGE {
ext.prefix = { "centrifuge_${meta.id}_combined_reports" }
publishDir = [
path: { "${params.outdir}/centrifuge/" },
mode: params.publish_dir_mode,
pattern: '*.{txt}'
]
}
2022-04-29 19:59:42 +00:00
withName: KAIJU_KAIJU {
2022-09-06 15:34:45 +00:00
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
publishDir = [
path: { "${params.outdir}/kaiju/${meta.db_name}/" },
mode: params.publish_dir_mode,
2022-04-29 19:59:42 +00:00
pattern: '*.tsv'
]
2022-04-29 19:59:42 +00:00
ext.args = { "${meta.db_params}" }
}
2022-04-29 20:02:44 +00:00
withName: KAIJU_KAIJU2TABLE {
2022-09-13 14:47:19 +00:00
ext.prefix = { "kaiju_${meta.id}_combined_reports" }
2022-04-10 04:44:10 +00:00
publishDir = [
2022-09-06 15:34:45 +00:00
path: { "${params.outdir}/kaiju/" },
2022-04-10 04:44:10 +00:00
mode: params.publish_dir_mode,
2022-04-29 19:59:42 +00:00
pattern: '*.{txt}'
2022-04-10 04:44:10 +00:00
]
}
withName: KAIJU_KAIJU2KRONA {
ext.args = '-v -u'
}
2022-04-29 20:02:44 +00:00
withName: DIAMOND_BLASTX {
2022-04-29 19:59:42 +00:00
ext.args = { "${meta.db_params}" }
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
2022-04-13 16:51:56 +00:00
publishDir = [
path: { "${params.outdir}/diamond/${meta.db_name}/" },
2022-04-13 16:51:56 +00:00
mode: params.publish_dir_mode,
pattern: '*.{blast,xml,txt,daa,sam,tsv,paf,log}'
2022-04-13 16:51:56 +00:00
]
}
2022-04-25 15:36:29 +00:00
2022-05-31 12:57:00 +00:00
withName: MOTUS_PROFILE {
2022-11-29 12:28:04 +00:00
ext.args = {
2022-11-29 12:31:00 +00:00
[
2022-12-01 12:42:21 +00:00
params.motus_remove_ncbi_ids ? "" : "-p",
params.motus_use_relative_abundance ? "" : "-c",
params.motus_save_mgc_read_counts ? "-M ${task.ext.prefix}.mgc" : ""
2022-11-29 12:28:04 +00:00
].join(',').replaceAll(','," ")
2022-11-29 12:31:00 +00:00
}
2022-07-12 09:39:26 +00:00
ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
2022-05-31 12:57:00 +00:00
publishDir = [
path: { "${params.outdir}/motus/${meta.db_name}/" },
2022-05-31 12:57:00 +00:00
mode: params.publish_dir_mode
]
}
2022-07-12 09:39:26 +00:00
withName: MOTUS_MERGE {
2022-09-16 06:36:34 +00:00
ext.args = { params.generate_biom_output ? "-B" : "" }
ext.prefix = { "motus_${meta.id}_combined_reports" }
2022-07-12 09:39:26 +00:00
publishDir = [
path: { "${params.outdir}/motus/" },
mode: params.publish_dir_mode
]
}
2022-04-29 19:59:42 +00:00
withName: CUSTOM_DUMPSOFTWAREVERSIONS {
2022-04-25 15:36:29 +00:00
publishDir = [
2022-04-29 19:59:42 +00:00
path: { "${params.outdir}/pipeline_info" },
2022-04-25 15:36:29 +00:00
mode: params.publish_dir_mode,
2022-04-29 19:59:42 +00:00
pattern: '*_versions.yml'
]
}
withName: MULTIQC {
publishDir = [
path: { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" },
mode: params.publish_dir_mode,
saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
2022-04-25 15:36:29 +00:00
]
}
}