kkonganti@0: // Define any required imports for this specific workflow
kkonganti@0: import java.nio.file.Paths
kkonganti@0: import nextflow.file.FileHelper
kkonganti@0: 
kkonganti@0: // Include any necessary methods
kkonganti@0: include { \
kkonganti@0:     summaryOfParams; stopNow; fastqEntryPointHelp; sendMail; \
kkonganti@0:     addPadding; wrapUpHelp   } from "${params.routines}"
kkonganti@0: include { kraken2Help        } from "${params.toolshelp}${params.fs}kraken2"
kkonganti@0: include { centrifugeHelp     } from "${params.toolshelp}${params.fs}centrifuge"
kkonganti@0: include { flyeHelp           } from "${params.toolshelp}${params.fs}flye"
kkonganti@0: include { serotypefinderHelp } from "${params.toolshelp}${params.fs}serotypefinder"
kkonganti@0: include { seqsero2Help       } from "${params.toolshelp}${params.fs}seqsero2"
kkonganti@0: include { mlstHelp           } from "${params.toolshelp}${params.fs}mlst"
kkonganti@0: include { abricateHelp       } from "${params.toolshelp}${params.fs}abricate"
kkonganti@0: 
kkonganti@0: // Exit if help requested before any subworkflows
kkonganti@0: if (params.help) {
kkonganti@0:     log.info help()
kkonganti@0:     exit 0
kkonganti@0: }
kkonganti@0: 
kkonganti@0: // Include any necessary modules and subworkflows
kkonganti@0: include { PROCESS_FASTQ           } from "${params.subworkflows}${params.fs}process_fastq"
kkonganti@0: include { FASTQC                  } from "${params.modules}${params.fs}fastqc${params.fs}main"
kkonganti@0: include { CENTRIFUGE_CLASSIFY     } from "${params.modules}${params.fs}centrifuge${params.fs}classify${params.fs}main"
kkonganti@0: include { CENTRIFUGE_PROCESS      } from "${params.modules}${params.fs}centrifuge${params.fs}process${params.fs}main"
kkonganti@0: include { SEQKIT_GREP             } from "${params.modules}${params.fs}seqkit${params.fs}grep${params.fs}main"
kkonganti@0: include { FLYE_ASSEMBLE           } from "${params.modules}${params.fs}flye${params.fs}assemble${params.fs}main"
kkonganti@0: include { KRAKEN2_CLASSIFY        } from "${params.modules}${params.fs}kraken2${params.fs}classify${params.fs}main"
kkonganti@0: include { KRAKEN2_EXTRACT_CONTIGS } from "${params.modules}${params.fs}kraken2${params.fs}extract_contigs${params.fs}main"
kkonganti@0: include { SEROTYPEFINDER          } from "${params.modules}${params.fs}serotypefinder${params.fs}main"
kkonganti@0: include { SEQSERO2                } from "${params.modules}${params.fs}seqsero2${params.fs}main"
kkonganti@0: include { MLST                    } from "${params.modules}${params.fs}mlst${params.fs}main"
kkonganti@0: include { ABRICATE_RUN            } from "${params.modules}${params.fs}abricate${params.fs}run${params.fs}main"
kkonganti@0: include { ABRICATE_SUMMARY        } from "${params.modules}${params.fs}abricate${params.fs}summary${params.fs}main"
kkonganti@0: include { TABLE_SUMMARY           } from "${params.modules}${params.fs}cat${params.fs}tables${params.fs}main"
kkonganti@0: include { MULTIQC                 } from "${params.modules}${params.fs}multiqc${params.fs}main"
kkonganti@0: include { DUMP_SOFTWARE_VERSIONS  } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main"
kkonganti@0: 
kkonganti@0: 
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     INPUTS AND ANY CHECKS FOR THE CENTRIFLAKEN WORKFLOW
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: def kraken2_db_dir = file ( "${params.kraken2_db}" )
kkonganti@0: def centrifuge_x = file ( "${params.centrifuge_x}" )
kkonganti@0: def reads_platform = 0
kkonganti@0: def abricate_dbs = [ 'ncbiamrplus', 'resfinder', 'megares', 'argannot' ]
kkonganti@0: 
kkonganti@0: reads_platform += (params.flye_nano_raw ? 1 : 0)
kkonganti@0: reads_platform += (params.flye_nano_corr ? 1 : 0)
kkonganti@0: reads_platform += (params.flye_nano_hq ? 1 : 0)
kkonganti@0: reads_platform += (params.flye_pacbio_raw ? 1 : 0)
kkonganti@0: reads_platform += (params.flye_pacbio_corr ? 1 : 0)
kkonganti@0: reads_platform += (params.flye_pacbio_hifi ? 1 : 0)
kkonganti@0: 
kkonganti@0: if (!kraken2_db_dir.exists() || !centrifuge_x.getParent().exists()) {
kkonganti@0:     stopNow("Please check if the following absolute paths are valid:\n" +
kkonganti@0:         "${params.kraken2_db}\n${params.centrifuge_x}\n" +
kkonganti@0:         "Cannot proceed further!")
kkonganti@0: }
kkonganti@0: 
kkonganti@0: if (reads_platform > 1 || reads_platform == 0) {
kkonganti@0:     msg_0 = (reads_platform > 1 ? "only" : "at least") 
kkonganti@0:     stopNow("Please mention ${msg_0} one read platform for use with the flye assembler\n" +
kkonganti@0:         "using any one of the following options:\n" +
kkonganti@0:         "--flye_nano_raw\n--flye_nano_corr\n--flye_nano_hq\n" +
kkonganti@0:         "--flye_pacbio_raw\n--flye_pacbio_corr\n--flye_pacbio_hifi")
kkonganti@0: }
kkonganti@0: 
kkonganti@0: if (params.centrifuge_extract_bug != params.kraken2_extract_bug) {
kkonganti@0:     stopNow("Please make sure that the bug to be extracted is same\n" +
kkonganti@0:     "for both --centrifuge_extract_bug and --kraken2_extract_bug options.")
kkonganti@0: }
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     RUN THE CENTRIFLAKEN WORKFLOW
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: workflow CENTRIFLAKEN {
kkonganti@0:     main:
kkonganti@0:         ch_asm_filtered_contigs = Channel.empty()
kkonganti@0:         ch_mqc_custom_tbl = Channel.empty()
kkonganti@0: 
kkonganti@0:         log.info summaryOfParams()
kkonganti@0: 
kkonganti@0:         PROCESS_FASTQ()
kkonganti@0:             .processed_reads
kkonganti@0:             .map {
kkonganti@0:                 meta, fastq ->
kkonganti@0:                 meta.centrifuge_x = params.centrifuge_x
kkonganti@0:                 meta.kraken2_db = params.kraken2_db
kkonganti@0:                 [meta, fastq]
kkonganti@0:             }
kkonganti@0:             .set { ch_processed_reads }
kkonganti@0:         
kkonganti@0:         PROCESS_FASTQ
kkonganti@0:             .out
kkonganti@0:             .versions
kkonganti@0:             .set { software_versions }
kkonganti@0: 
kkonganti@0:         FASTQC ( ch_processed_reads )
kkonganti@0: 
kkonganti@47:         CENTRIFUGE_CLASSIFY ( ch_processed_reads )
kkonganti@0: 
kkonganti@47:         CENTRIFUGE_PROCESS (
kkonganti@0:             CENTRIFUGE_CLASSIFY.out.report
kkonganti@0:                 .join( CENTRIFUGE_CLASSIFY.out.output )
kkonganti@0:         )
kkonganti@0: 
kkonganti@47:         ch_processed_reads.join ( CENTRIFUGE_PROCESS.out.extracted )
kkonganti@0:             .set { ch_centrifuge_extracted }
kkonganti@0: 
kkonganti@0:         SEQKIT_GREP ( ch_centrifuge_extracted )
kkonganti@0: 
kkonganti@0:         FLYE_ASSEMBLE ( SEQKIT_GREP.out.fastx )
kkonganti@0: 
kkonganti@0:         FLYE_ASSEMBLE
kkonganti@0:             .out
kkonganti@0:             .assembly
kkonganti@0:             .set { ch_flye_assembly }
kkonganti@0: 
kkonganti@0:         ch_flye_assembly.ifEmpty { [ false, false ] }
kkonganti@0: 
kkonganti@47:         KRAKEN2_CLASSIFY ( ch_flye_assembly )
kkonganti@0: 
kkonganti@0:         KRAKEN2_EXTRACT_CONTIGS (
kkonganti@0:             ch_flye_assembly
kkonganti@0:                 .join( KRAKEN2_CLASSIFY.out.kraken_output ),
kkonganti@0:             params.kraken2_extract_bug
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         KRAKEN2_EXTRACT_CONTIGS
kkonganti@0:             .out
kkonganti@0:             .asm_filtered_contigs
kkonganti@0:             .map {
kkonganti@0:                 meta, fastq ->
kkonganti@0:                 meta.organism = params.kraken2_extract_bug.split(/\s+/)[0].capitalize()
kkonganti@0:                 meta.serotypefinder_db = params.serotypefinder_db
kkonganti@0:                 [meta, fastq]
kkonganti@0:             }
kkonganti@0:             .set { ch_asm_filtered_contigs }
kkonganti@0: 
kkonganti@0:         SEROTYPEFINDER ( ch_asm_filtered_contigs )
kkonganti@0: 
kkonganti@0:         SEQSERO2 ( ch_asm_filtered_contigs )
kkonganti@0: 
kkonganti@0:         MLST ( ch_asm_filtered_contigs )
kkonganti@0: 
kkonganti@0:         ABRICATE_RUN ( 
kkonganti@0:             ch_asm_filtered_contigs, 
kkonganti@0:             abricate_dbs
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         ABRICATE_RUN
kkonganti@0:             .out
kkonganti@0:             .abricated
kkonganti@0:             .map { meta, abres -> [ abricate_dbs, abres ] }
kkonganti@0:             .groupTuple(by: [0])
kkonganti@0:             .map { it -> tuple ( it[0], it[1].flatten() ) }
kkonganti@0:             .set { ch_abricated }
kkonganti@0: 
kkonganti@0:         ABRICATE_SUMMARY ( ch_abricated )
kkonganti@0: 
kkonganti@0:         // ABRICATE_SUMMARY.out.ecoli_vf.set { ch_abricate_summary_ecoli_vf }
kkonganti@0:         // ch_abricate_summary_ecoli_vf.ifEmpty { [ false, false ] }
kkonganti@0: 
kkonganti@0:         CENTRIFUGE_CLASSIFY.out.kreport
kkonganti@0:             .map { meta, kreport -> [ kreport ] }
kkonganti@0:             .flatten()
kkonganti@0:             .concat ( 
kkonganti@0:                 KRAKEN2_CLASSIFY.out.kraken_report
kkonganti@0:                 .map { meta, kreport -> [ kreport ] }
kkonganti@0:                 .flatten(),
kkonganti@0:                 FASTQC.out.zip
kkonganti@0:                 .map { meta, zip -> [ zip ] }
kkonganti@0:                 .flatten()
kkonganti@0:             )
kkonganti@0:             .set { ch_mqc_classify }
kkonganti@0: 
kkonganti@0:         if (params.serotypefinder_run) {
kkonganti@0:             SEROTYPEFINDER
kkonganti@0:                 .out
kkonganti@0:                 .serotyped
kkonganti@0:                 .map { meta, tsv -> [ 'serotypefinder', tsv ] }
kkonganti@0:                 .groupTuple(by: [0])
kkonganti@0:                 .map { it -> tuple ( it[0], it[1].flatten() ) }
kkonganti@0:                 .set { ch_mqc_custom_tbl }
kkonganti@0:         } else if (params.seqsero2_run) {
kkonganti@0:             SEQSERO2
kkonganti@0:                 .out
kkonganti@0:                 .serotyped
kkonganti@0:                 .map { meta, tsv -> [ 'seqsero2', tsv ] }
kkonganti@0:                 .groupTuple(by: [0])
kkonganti@0:                 .map { it -> tuple ( it[0], it[1].flatten() ) }
kkonganti@0:                 .set { ch_mqc_custom_tbl }
kkonganti@0:         }
kkonganti@0: 
kkonganti@0:         ch_mqc_custom_tbl
kkonganti@0:             .concat (
kkonganti@0:                 ABRICATE_SUMMARY.out.ncbiamrplus.map{ it -> tuple ( it[0], it[1] )},
kkonganti@0:                 ABRICATE_SUMMARY.out.resfinder.map{ it -> tuple ( it[0], it[1] )},
kkonganti@0:                 ABRICATE_SUMMARY.out.megares.map{ it -> tuple ( it[0], it[1] )},
kkonganti@0:                 ABRICATE_SUMMARY.out.argannot.map{ it -> tuple ( it[0], it[1] )},
kkonganti@0:             )
kkonganti@0:             .groupTuple(by: [0])
kkonganti@0:             .map { it -> [ it[0], it[1].flatten() ]}
kkonganti@0:             .set { ch_mqc_custom_tbl }
kkonganti@0: 
kkonganti@47:         TABLE_SUMMARY ( ch_mqc_custom_tbl )
kkonganti@0: 
kkonganti@0:         DUMP_SOFTWARE_VERSIONS (
kkonganti@0:             software_versions
kkonganti@0:                 .mix (
kkonganti@0:                     FASTQC.out.versions,
kkonganti@0:                     CENTRIFUGE_CLASSIFY.out.versions,
kkonganti@0:                     CENTRIFUGE_PROCESS.out.versions,
kkonganti@0:                     SEQKIT_GREP.out.versions,
kkonganti@0:                     FLYE_ASSEMBLE.out.versions.ifEmpty(null),
kkonganti@0:                     KRAKEN2_CLASSIFY.out.versions.ifEmpty(null),
kkonganti@0:                     KRAKEN2_EXTRACT_CONTIGS.out.versions.ifEmpty(null),
kkonganti@0:                     SEROTYPEFINDER.out.versions.ifEmpty(null),
kkonganti@0:                     SEQSERO2.out.versions.ifEmpty(null),
kkonganti@0:                     MLST.out.versions.ifEmpty(null),
kkonganti@0:                     ABRICATE_RUN.out.versions.ifEmpty(null),
kkonganti@0:                     ABRICATE_SUMMARY.out.versions.ifEmpty(null),
kkonganti@0:                     TABLE_SUMMARY.out.versions.ifEmpty(null)
kkonganti@0:                 )
kkonganti@0:                 .unique()
kkonganti@0:                 .collectFile(name: 'collected_versions.yml')
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         DUMP_SOFTWARE_VERSIONS
kkonganti@0:             .out
kkonganti@0:             .mqc_yml
kkonganti@0:             .concat (
kkonganti@0:                 ch_mqc_classify,
kkonganti@0:                 TABLE_SUMMARY.out.mqc_yml
kkonganti@0:             )
kkonganti@0:             .collect()
kkonganti@0:             .set { ch_multiqc }
kkonganti@0: 
kkonganti@47:         MULTIQC ( ch_multiqc )
kkonganti@0: }
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     ON COMPLETE, SHOW GORY DETAILS OF ALL PARAMS WHICH WILL BE HELPFUL TO DEBUG
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: workflow.onComplete {
kkonganti@0:     if (workflow.success) {
kkonganti@0:         // CREATE APPROPRIATE DIRECTORIES AND MOVE AS REQUESTED BY STAKEHOLDER(S)
kkonganti@0:         //
kkonganti@0:         // Nextflow's .moveTo will error out if directories contain files and it
kkonganti@0:         // would be complex to include logic to skip directories 
kkonganti@0:         //
kkonganti@0:         def final_intermediate_dir = "${params.output}${params.fs}${params.pipeline}-steps"
kkonganti@0:         def final_results_dir = "${params.output}${params.fs}${params.pipeline}-results"
kkonganti@0:         def kraken2_ext_contigs =  file( "${final_intermediate_dir}${params.fs}kraken2_extract_contigs", type: 'dir' )
kkonganti@0:         def final_intermediate = file( final_intermediate_dir, type: 'dir' )
kkonganti@0:         def final_results = file( final_results_dir, type: 'dir' )
kkonganti@0:         def pipeline_output = file( params.output, type: 'dir' )
kkonganti@0: 
kkonganti@0:         if ( !final_intermediate.exists() ) {
kkonganti@0:             final_intermediate.mkdirs()
kkonganti@0: 
kkonganti@0:             FileHelper.visitFiles(Paths.get("${params.output}"), '*') {
kkonganti@0:                 if ( !(it.name ==~ /^(${params.cfsanpipename}|multiqc|\.nextflow|${workflow.workDir.name}|${params.pipeline}).*/) ) {
kkonganti@0:                     FileHelper.movePath(
kkonganti@0:                         it, Paths.get( "${final_intermediate_dir}${params.fs}${it.name}" )
kkonganti@0:                     )
kkonganti@0:                 }
kkonganti@0:             }
kkonganti@0:         }
kkonganti@0: 
kkonganti@0:         if ( kraken2_ext_contigs.exists() && !final_results.exists() ) {
kkonganti@0:             final_results.mkdirs()
kkonganti@0: 
kkonganti@0:             FileHelper.movePath(
kkonganti@0:                 Paths.get( "${final_intermediate_dir}${params.fs}kraken2_extract_contigs" ),
kkonganti@0:                 Paths.get( "${final_results_dir}${params.fs}kraken2_extract_contigs" )
kkonganti@0:             )
kkonganti@0:         }
kkonganti@0: 
kkonganti@0:         sendMail()
kkonganti@0:     }
kkonganti@0: }
kkonganti@0: 
kkonganti@0: workflow.onError {
kkonganti@0:     sendMail()
kkonganti@0: }
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     HELPER METHODS FOR CENTRIFLAKEN WORKFLOW
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: def help() {
kkonganti@0: 
kkonganti@0:     Map helptext = [:]
kkonganti@0: 
kkonganti@47:     helptext.putAll (
kkonganti@0:         fastqEntryPointHelp() +
kkonganti@0:         kraken2Help(params).text +
kkonganti@0:         centrifugeHelp(params).text +
kkonganti@0:         flyeHelp(params).text +
kkonganti@0:         serotypefinderHelp(params).text +
kkonganti@0:         seqsero2Help(params).text +
kkonganti@0:         mlstHelp(params).text +
kkonganti@0:         abricateHelp(params).text +
kkonganti@0:         wrapUpHelp()
kkonganti@0:     )
kkonganti@0: 
kkonganti@0:     return addPadding(helptext)
kkonganti@0: }