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 "${params.routines}"
kkonganti@0: include { dpubmlstpyHelp     } from "${params.toolshelp}${params.fs}dpubmlstpy"
kkonganti@0: include { checkm2predictHelp } from "${params.toolshelp}${params.fs}checkm2predict"
kkonganti@0: include { guncrunHelp        } from "${params.toolshelp}${params.fs}guncrun"
kkonganti@0: include { mlstHelp           } from "${params.toolshelp}${params.fs}mlst"
kkonganti@0: 
kkonganti@0: // Exit if help requested before any subworkflows
kkonganti@0: if (params.help) {
kkonganti@0:     log.info help()
kkonganti@0:     exit 0
kkonganti@0: }
kkonganti@0: 
kkonganti@0: // Include any necessary modules and subworkflows
kkonganti@0: include { DOWNLOAD_PDG_METADATA    } from "${params.modules}${params.fs}download_pdg_metadata${params.fs}main"
kkonganti@0: include { DOWNLOAD_PUBMLST_SCHEME  } from "${params.modules}${params.fs}download_pubmlst_scheme${params.fs}main"
kkonganti@0: include { FILTER_PDG_METADATA      } from "${params.modules}${params.fs}filter_pdg_metadata${params.fs}main"
kkonganti@0: include { GUNC_RUN                 } from "${params.modules}${params.fs}gunc${params.fs}run${params.fs}main"
kkonganti@0: include { CHECKM2_PREDICT          } from "${params.modules}${params.fs}checkm2${params.fs}predict${params.fs}main"
kkonganti@0: include { QUAL_PASSED_GENOMES      } from "${params.modules}${params.fs}custom${params.fs}qual_passed_genomes${params.fs}main"
kkonganti@0: include { SCAFFOLD_GENOMES         } from "${params.modules}${params.fs}scaffold_genomes${params.fs}main"
kkonganti@0: include { MLST                     } from "${params.modules}${params.fs}mlst${params.fs}main"
kkonganti@0: include { INDEX_PDG_METADATA       } from "${params.modules}${params.fs}index_pdg_metadata${params.fs}main"
kkonganti@0: include { MASH_SKETCH              } from "${params.modules}${params.fs}mash${params.fs}sketch${params.fs}main"
kkonganti@0: include { MASH_PASTE               } from "${params.modules}${params.fs}mash${params.fs}paste${params.fs}main"
kkonganti@0: include { DUMP_SOFTWARE_VERSIONS   } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main"
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     INPUTS AND ANY CHECKS FOR THE CRONOLOGY_DB WORKFLOW
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: if (!params.output) {
kkonganti@0:     stopNow("Please mention the absolute UNIX path to store the DB flat files\n" +
kkonganti@0:             "using the --output option.\n" +
kkonganti@0:         "Ex: --output /path/to/cronology/db_files")
kkonganti@0: }
kkonganti@0: 
kkonganti@0: checkDBPathExists(params.guncrun_dbpath, 'GUNC')
kkonganti@0: checkDBPathExists(params.checkm2predict_dbpath, 'CheckM2')
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     RUN THE CRONOLOGY_DB WORKFLOW
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: workflow CRONOLOGY_DB {
kkonganti@0:     main:
kkonganti@0:         log.info summaryOfParams()
kkonganti@0: 
kkonganti@0:         DOWNLOAD_PDG_METADATA ( params.pdg_release ?: null )
kkonganti@0: 
kkonganti@0:         DOWNLOAD_PDG_METADATA.out.versions
kkonganti@0:             .set { software_versions }
kkonganti@0: 
kkonganti@0:         DOWNLOAD_PUBMLST_SCHEME ( params.dpubmlstpy_org ?: null )
kkonganti@0: 
kkonganti@0:         FILTER_PDG_METADATA (
kkonganti@0:             DOWNLOAD_PDG_METADATA.out.accs
kkonganti@0:                 .splitText(by: params.genomes_chunk, file: true)
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         FILTER_PDG_METADATA.out.accs_chunk_tbl
kkonganti@0:             .collectFile(sort: { acc_f -> acc_f.simpleName })
kkonganti@0:             .multiMap { acc_chunk_file ->
kkonganti@0:                 def meta = [:]
kkonganti@0:                 meta.id = 'AssemblyQC'
kkonganti@0:                 meta.phone_ncbi = true
kkonganti@0:                 gunc: [ meta, params.guncrun_dbpath, acc_chunk_file ]
kkonganti@0:                 checkm2: [ meta, params.checkm2predict_dbpath, acc_chunk_file ]
kkonganti@0:             }
kkonganti@0:             .set { ch_run_qual_on_these_accs }
kkonganti@0: 
kkonganti@0:         CHECKM2_PREDICT ( ch_run_qual_on_these_accs.checkm2 )
kkonganti@0: 
kkonganti@0:         GUNC_RUN ( ch_run_qual_on_these_accs.gunc )
kkonganti@0: 
kkonganti@0:         QUAL_PASSED_GENOMES (
kkonganti@0:             CHECKM2_PREDICT.out.quality_report_passed
kkonganti@0:                 .map { meta, qual ->
kkonganti@0:                     [ qual ]
kkonganti@0:                 }
kkonganti@0:                 .collect()
kkonganti@0:                 .flatten()
kkonganti@0:                 .collectFile(name: 'checkm2_quality_passed.txt'),
kkonganti@0:             GUNC_RUN.out.quality_report_passed
kkonganti@0:                 .map { meta, qual ->
kkonganti@0:                     [ qual ]
kkonganti@0:                 }
kkonganti@0:                 .collect()
kkonganti@0:                 .flatten()
kkonganti@0:                 .collectFile(name: 'gunc_quality_passed.txt')
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         SCAFFOLD_GENOMES (
kkonganti@0:             QUAL_PASSED_GENOMES.out.accs
kkonganti@0:                 .splitText(by: params.genomes_chunk, file: true)
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         SCAFFOLD_GENOMES.out.scaffolded
kkonganti@0:             .multiMap { scaffolded ->
kkonganti@0:                 def meta = [:]
kkonganti@0:                 meta.id = (params.pdg_release ?: 'NCBI Pathogen Genomes')
kkonganti@0:                 mlst: [ meta, scaffolded ]
kkonganti@0:                 mash: [ meta, scaffolded ]
kkonganti@0:             }
kkonganti@0:             .set { ch_scaffolded_genomes }
kkonganti@0: 
kkonganti@0:         MLST (
kkonganti@0:             ch_scaffolded_genomes.mlst
kkonganti@0:                 .combine( DOWNLOAD_PUBMLST_SCHEME.out.pubmlst_dir )
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         MLST.out.tsv
kkonganti@0:             .map { meta, tsv -> 
kkonganti@0:                 tsv
kkonganti@0:             }
kkonganti@0:             .collectFile(
kkonganti@0:                 name: 'mlst_results.tsv',
kkonganti@0:                 keepHeader: true,
kkonganti@0:                 skip: 1
kkonganti@0:             )
kkonganti@0:             .set { ch_mlst_results }
kkonganti@0: 
kkonganti@0:         INDEX_PDG_METADATA (
kkonganti@0:             DOWNLOAD_PDG_METADATA.out.pdg_metadata,
kkonganti@0:             DOWNLOAD_PDG_METADATA.out.snp_cluster_metadata,
kkonganti@0:             DOWNLOAD_PDG_METADATA.out.accs,
kkonganti@0:             ch_mlst_results
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         MASH_SKETCH ( 
kkonganti@0:             ch_scaffolded_genomes.mash
kkonganti@0:                 .map { it -> tuple ( it[0], it[1].flatten() ) }
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         MASH_PASTE (
kkonganti@0:             MASH_SKETCH.out.sketch
kkonganti@0:                 .map { meta, sketch ->
kkonganti@0:                     [ [id: (params.pdg_release ?: 'NCBI Pathogen Genomes')], sketch ]
kkonganti@0:                 }
kkonganti@0:                 .groupTuple(by: [0])
kkonganti@0:         )
kkonganti@0: 
kkonganti@0:         DUMP_SOFTWARE_VERSIONS (
kkonganti@0:             software_versions
kkonganti@0:                 .mix (
kkonganti@0:                     DOWNLOAD_PDG_METADATA.out.versions,
kkonganti@0:                     DOWNLOAD_PUBMLST_SCHEME.out.versions,
kkonganti@0:                     FILTER_PDG_METADATA.out.versions,
kkonganti@0:                     CHECKM2_PREDICT.out.versions,
kkonganti@0:                     GUNC_RUN.out.versions,
kkonganti@0:                     QUAL_PASSED_GENOMES.out.versions,
kkonganti@0:                     SCAFFOLD_GENOMES.out.versions,
kkonganti@0:                     MLST.out.versions,
kkonganti@0:                     INDEX_PDG_METADATA.out.versions,
kkonganti@0:                     MASH_SKETCH.out.versions,
kkonganti@0:                     MASH_PASTE.out.versions
kkonganti@0:                 )
kkonganti@0:                 .unique()
kkonganti@0:                 .collectFile(name: 'collected_versions.yml')
kkonganti@0:         )
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:         sendMail()
kkonganti@0:     }
kkonganti@0: }
kkonganti@0: 
kkonganti@0: workflow.onError {
kkonganti@0:     sendMail()
kkonganti@0: }
kkonganti@0: 
kkonganti@0: /*
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     METHOD TO CHECK IF DB PATHS EXIST
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: def checkDBPathExists(db_path, msg) {
kkonganti@0:     db_path_obj = file( db_path )
kkonganti@0: 
kkonganti@0:     if (!db_path_obj.exists()) {
kkonganti@0:         stopNow("Please check if the database path for ${msg}\n" +
kkonganti@0:             "[ ${db_path} ]\nexists.")
kkonganti@0:     }
kkonganti@0: }/*
kkonganti@0: 
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0:     HELP TEXT METHODS FOR CRONOLOGY WORKFLOW
kkonganti@0: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@0: */
kkonganti@0: 
kkonganti@0: def help() {
kkonganti@0: 
kkonganti@0:     Map helptext = [:]
kkonganti@0: 
kkonganti@0:     helptext.putAll (
kkonganti@0:         fastqEntryPointHelp().findAll {
kkonganti@0:             it.key =~ /Required|output|Other|Workflow|Author|Version/
kkonganti@0:         } +
kkonganti@0:         dpubmlstpyHelp(params).text +
kkonganti@0:         checkm2predictHelp(params).text +
kkonganti@0:         guncrunHelp(params).text +
kkonganti@0:         wrapUpHelp()
kkonganti@0:     )
kkonganti@0: 
kkonganti@0:     return addPadding(helptext)
kkonganti@0: }