1
0
Fork 0
mirror of https://github.com/MillironX/taxprofiler.git synced 2024-11-29 08:49:54 +00:00
taxprofiler/workflows/taxprofiler.nf
James Fellows Yates aa2d07c42a Fix merge cockup
2022-04-13 14:19:19 +02:00

261 lines
10 KiB
Text

/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
VALIDATE INPUTS
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/
def summary_params = NfcoreSchema.paramsSummaryMap(workflow, params)
// Validate input parameters
WorkflowTaxprofiler.initialise(params, log)
// TODO nf-core: Add all file path parameters for the pipeline to the list below
// Check input path parameters to see if they exist
def checkPathParamList = [ params.input, params.databases, params.shortread_hostremoval_reference,
params.shortread_hostremoval_index, params.multiqc_config
]
for (param in checkPathParamList) { if (param) { file(param, checkIfExists: true) } }
// Check mandatory parameters
if (params.input ) { ch_input = file(params.input) } else { exit 1, 'Input samplesheet not specified!' }
if (params.databases) { ch_databases = file(params.databases) } else { exit 1, 'Input database sheet not specified!' }
if (params.shortread_clipmerge_mergepairs && params.run_malt ) log.warn "[nf-core/taxprofiler] warning: MALT does not accept uncollapsed paired-reads. Pairs will be profiled as separate files."
if (params.shortread_clipmerge_excludeunmerged && !params.shortread_clipmerge_mergepairs) exit 1, "[nf-core/taxprofiler] error: cannot include unmerged reads when merging not turned on. Please specify --shortread_clipmerge_mergepairs"
if (params.perform_shortread_hostremoval && !params.shortread_hostremoval_reference) { exit 1, "[nf-core/taxprofiler] error: --shortread_hostremoval requested but no --shortread_hostremoval_reference FASTA supplied. Check input." }
if (!params.shortread_hostremoval_reference && params.shortread_hostremoval_reference_index) { exit 1, "[nf-core/taxprofiler] error: --shortread_hostremoval_index provided but no --shortread_hostremoval_reference FASTA supplied. Check input." }
if (params.shortread_hostremoval_reference ) { ch_reference = file(params.shortread_hostremoval_reference) }
if (params.shortread_hostremoval_index ) { ch_reference_index = file(params.shortread_hostremoval_index ) } else { ch_reference_index = [] }
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
CONFIG FILES
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/
ch_multiqc_config = file("$projectDir/assets/multiqc_config.yml", checkIfExists: true)
ch_multiqc_custom_config = params.multiqc_config ? Channel.fromPath(params.multiqc_config) : Channel.empty()
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
IMPORT LOCAL MODULES/SUBWORKFLOWS
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/
//
// SUBWORKFLOW: Consisting of a mix of local and nf-core/modules
//
include { INPUT_CHECK } from '../subworkflows/local/input_check'
include { DB_CHECK } from '../subworkflows/local/db_check'
include { SHORTREAD_PREPROCESSING } from '../subworkflows/local/shortread_preprocessing'
include { LONGREAD_PREPROCESSING } from '../subworkflows/local/longread_preprocessing'
include { SHORTREAD_HOSTREMOVAL } from '../subworkflows/local/shortread_hostremoval'
include { SHORTREAD_COMPLEXITYFILTERING } from '../subworkflows/local/shortread_complexityfiltering'
include { PROFILING } from '../subworkflows/local/profiling'
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
IMPORT NF-CORE MODULES/SUBWORKFLOWS
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/
//
// MODULE: Installed directly from nf-core/modules
//
include { FASTQC } from '../modules/nf-core/modules/fastqc/main'
include { MULTIQC } from '../modules/nf-core/modules/multiqc/main'
include { CUSTOM_DUMPSOFTWAREVERSIONS } from '../modules/nf-core/modules/custom/dumpsoftwareversions/main'
include { CAT_FASTQ } from '../modules/nf-core/modules/cat/fastq/main'
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
RUN MAIN WORKFLOW
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/
// Info required for completion email and summary
def multiqc_report = []
workflow TAXPROFILER {
ch_versions = Channel.empty()
/*
SUBWORKFLOW: Read in samplesheet, validate and stage input files
*/
INPUT_CHECK (
ch_input
)
ch_versions = ch_versions.mix(INPUT_CHECK.out.versions)
DB_CHECK (
ch_databases
)
ch_versions = ch_versions.mix(DB_CHECK.out.versions)
/*
MODULE: Run FastQC
*/
ch_input_for_fastqc = INPUT_CHECK.out.fastq.mix( INPUT_CHECK.out.nanopore )
FASTQC (
ch_input_for_fastqc
)
ch_versions = ch_versions.mix(FASTQC.out.versions.first())
/*
SUBWORKFLOW: PERFORM PREPROCESSING
*/
if ( params.perform_shortread_clipmerge ) {
ch_shortreads_preprocessed = SHORTREAD_PREPROCESSING ( INPUT_CHECK.out.fastq ).reads
} else {
ch_shortreads_preprocessed = INPUT_CHECK.out.fastq
}
if ( params.perform_longread_clip ) {
ch_longreads_preprocessed = LONGREAD_PREPROCESSING ( INPUT_CHECK.out.nanopore ).reads
.map { it -> [ it[0], [it[1]] ] }
ch_versions = ch_versions.mix(LONGREAD_PREPROCESSING.out.versions.first())
} else {
ch_longreads_preprocessed = INPUT_CHECK.out.nanopore
}
/*
SUBWORKFLOW: COMPLEXITY FILTERING
*/
if ( params.perform_shortread_complexityfilter ) {
ch_shortreads_filtered = SHORTREAD_COMPLEXITYFILTERING ( ch_shortreads_preprocessed ).reads
} else {
ch_shortreads_filtered = ch_shortreads_preprocessed
}
/*
SUBWORKFLOW: HOST REMOVAL
*/
if ( params.perform_shortread_hostremoval ) {
ch_shortreads_hostremoved = SHORTREAD_HOSTREMOVAL ( ch_shortreads_filtered, ch_reference, ch_reference_index ).reads
ch_versions = ch_versions.mix(SHORTREAD_HOSTREMOVAL.out.versions.first())
} else {
ch_shortreads_hostremoved = ch_shortreads_filtered
}
if ( params.perform_runmerging ) {
ch_reads_for_cat_branch = ch_shortreads_hostremoved
.mix( ch_longreads_preprocessed )
.map {
meta, reads ->
def meta_new = meta.clone()
meta_new.remove('run_accession')
[ meta_new, reads ]
}
.groupTuple()
.map {
meta, reads ->
[ meta, reads.flatten() ]
}
.branch {
meta, reads ->
// we can't concatenate files if there is not a second run, we branch
// here to separate them out, and mix back in after for efficiency
cat: ( meta.single_end && reads.size() > 1 ) || ( !meta.single_end && reads.size() > 2 )
skip: true
}
ch_reads_runmerged = CAT_FASTQ ( ch_reads_for_cat_branch.cat ).reads
.mix( ch_reads_for_cat_branch.skip )
.map {
meta, reads ->
[ meta, [ reads ].flatten() ]
}
} else {
ch_reads_runmerged = ch_shortreads_hostremoved
.mix( ch_longreads_preprocessed )
}
/*
SUBWORKFLOW: PROFILING
*/
PROFILING ( ch_reads_runmerged, DB_CHECK.out.dbs )
ch_versions = ch_versions.mix( PROFILING.out.versions )
/*
MODULE: MultiQC
*/
CUSTOM_DUMPSOFTWAREVERSIONS (
ch_versions.unique().collectFile(name: 'collated_versions.yml')
)
workflow_summary = WorkflowTaxprofiler.paramsSummaryMultiqc(workflow, summary_params)
ch_workflow_summary = Channel.value(workflow_summary)
ch_multiqc_files = Channel.empty()
ch_multiqc_files = ch_multiqc_files.mix(Channel.from(ch_multiqc_config))
ch_multiqc_files = ch_multiqc_files.mix(ch_multiqc_custom_config.collect().ifEmpty([]))
ch_multiqc_files = ch_multiqc_files.mix(ch_workflow_summary.collectFile(name: 'workflow_summary_mqc.yaml'))
ch_multiqc_files = ch_multiqc_files.mix(CUSTOM_DUMPSOFTWAREVERSIONS.out.mqc_yml.collect())
ch_multiqc_files = ch_multiqc_files.mix(FASTQC.out.zip.collect{it[1]}.ifEmpty([]))
if (params.perform_shortread_clipmerge) {
ch_multiqc_files = ch_multiqc_files.mix( SHORTREAD_PREPROCESSING.out.mqc.collect{it[1]}.ifEmpty([]) )
ch_versions = ch_versions.mix( SHORTREAD_PREPROCESSING.out.versions )
}
if (params.perform_longread_clip) {
ch_multiqc_files = ch_multiqc_files.mix( LONGREAD_PREPROCESSING.out.mqc.collect{it[1]}.ifEmpty([]) )
ch_versions = ch_versions.mix( LONGREAD_PREPROCESSING.out.versions )
}
if (params.perform_shortread_complexityfilter){
ch_multiqc_files = ch_multiqc_files.mix( SHORTREAD_COMPLEXITYFILTERING.out.mqc.collect{it[1]}.ifEmpty([]) )
ch_versions = ch_versions.mix( SHORTREAD_COMPLEXITYFILTERING.out.versions )
}
if (params.perform_shortread_hostremoval) {
ch_multiqc_files = ch_multiqc_files.mix(SHORTREAD_HOSTREMOVAL.out.mqc.collect{it[1]}.ifEmpty([]))
ch_versions = ch_versions.mix(SHORTREAD_HOSTREMOVAL.out.versions)
}
if (params.perform_runmerging){
ch_versions = ch_versions.mix(CAT_FASTQ.out.versions)
}
ch_multiqc_files = ch_multiqc_files.mix( PROFILING.out.mqc )
// TODO create multiQC module for metaphlan
MULTIQC (
ch_multiqc_files.collect()
)
multiqc_report = MULTIQC.out.report.toList()
ch_versions = ch_versions.mix(MULTIQC.out.versions)
}
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
COMPLETION EMAIL AND SUMMARY
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/
workflow.onComplete {
if (params.email || params.email_on_fail) {
NfcoreTemplate.email(workflow, params, summary_params, projectDir, log, multiqc_report)
}
NfcoreTemplate.summary(workflow, params, log)
}
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
THE END
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*/