kkonganti@1: // Define any required imports for this specific workflow
kkonganti@1: import java.nio.file.Paths
kkonganti@1: import nextflow.file.FileHelper
kkonganti@1: 
kkonganti@1: 
kkonganti@1: // Include any necessary methods
kkonganti@1: include { \
kkonganti@1:     fastqEntryPointHelp; summaryOfParams; stopNow; sendMail; \
kkonganti@1:     addPadding; wrapUpHelp    } from "${params.routines}"
kkonganti@1: include { wcompHelp           } from "${params.toolshelp}${params.fs}wcomp"
kkonganti@1: include { wsnpHelp            } from "${params.toolshelp}${params.fs}wsnp"
kkonganti@1: include { mashsketchHelp      } from "${params.toolshelp}${params.fs}mashsketch"
kkonganti@1: 
kkonganti@1: 
kkonganti@1: // Exit if help requested before any subworkflows
kkonganti@1: if (params.help) {
kkonganti@1:     log.info help()
kkonganti@1:     exit 0
kkonganti@1: }
kkonganti@1: 
kkonganti@1: 
kkonganti@1: // Include any necessary modules and subworkflows
kkonganti@1: include { DOWNLOAD_PDG_METADATA    } from "${params.modules}${params.fs}download_pdg_metadata${params.fs}main"
kkonganti@1: include { FILTER_PDG_METADATA      } from "${params.modules}${params.fs}filter_pdg_metadata${params.fs}main"
kkonganti@1: include { DB_PER_COMPUTED_SEROTYPE } from "${params.modules}${params.fs}db_per_computed_serotype${params.fs}main"
kkonganti@1: include { DB_PER_SNP_CLUSTER       } from "${params.modules}${params.fs}db_per_snp_cluster${params.fs}main"
kkonganti@1: include { INDEX_METADATA           } from "${params.modules}${params.fs}index_metadata${params.fs}main"
kkonganti@1: include { SCAFFOLD_GENOMES         } from "${params.modules}${params.fs}scaffold_genomes${params.fs}main"
kkonganti@1: include { MASH_SKETCH              } from "${params.modules}${params.fs}mash${params.fs}sketch${params.fs}main"
kkonganti@1: include { DUMP_SOFTWARE_VERSIONS   } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main"
kkonganti@1: 
kkonganti@1: /*
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1:     INPUTS AND ANY CHECKS FOR THE BETTERCALLSAL_DB WORKFLOW
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1: */
kkonganti@1: 
kkonganti@1: if (!params.output) {
kkonganti@1:     stopNow("Please mention the absolute UNIX path to store the DB flat files\n" +
kkonganti@1:             "using the --output option.\n" +
kkonganti@1:         "Ex: --output /path/to/bettercallsal/db_files")
kkonganti@1: }
kkonganti@1: 
kkonganti@1: /*
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1:     RUN THE BETTERCALLSAL_DB WORKFLOW
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1: */
kkonganti@1: 
kkonganti@1: workflow BETTERCALLSAL_DB {
kkonganti@1:     main:
kkonganti@1:         log.info summaryOfParams()
kkonganti@1: 
kkonganti@1:         DOWNLOAD_PDG_METADATA ( params.pdg_release ?: null )
kkonganti@1: 
kkonganti@1:         DOWNLOAD_PDG_METADATA
kkonganti@1:             .out
kkonganti@1:             .versions
kkonganti@1:             .set { software_versions }
kkonganti@1: 
kkonganti@1:         FILTER_PDG_METADATA ( DOWNLOAD_PDG_METADATA.out.accs )
kkonganti@1: 
kkonganti@1:         DB_PER_COMPUTED_SEROTYPE (
kkonganti@1:             FILTER_PDG_METADATA.out.accs_chunk_tbl,
kkonganti@1:             DOWNLOAD_PDG_METADATA.out.pdg_metadata
kkonganti@1:         )
kkonganti@1: 
kkonganti@1:         DB_PER_SNP_CLUSTER (
kkonganti@1:             FILTER_PDG_METADATA.out.accs_chunk_tbl,
kkonganti@1:             DOWNLOAD_PDG_METADATA.out.pdg_metadata,
kkonganti@1:             DOWNLOAD_PDG_METADATA.out.snp_cluster_metadata
kkonganti@1:         )
kkonganti@1: 
kkonganti@1:         DB_PER_COMPUTED_SEROTYPE.out.genome_paths
kkonganti@1:             .map { query ->
kkonganti@1:                 kv = [:]
kkonganti@1:                 kv['id'] = 'comp'
kkonganti@1:                 [ kv, query ]
kkonganti@1:             }
kkonganti@1:             .concat(
kkonganti@1:                 DB_PER_SNP_CLUSTER.out.genome_paths
kkonganti@1:                     .map { query ->
kkonganti@1:                         kv = [:]
kkonganti@1:                         kv['id'] = 'snp'
kkonganti@1:                         [ kv, query ]
kkonganti@1:                     }
kkonganti@1:             )
kkonganti@1:             .groupTuple(by: [0])
kkonganti@1:             .set { ch_mash_these_genomes }
kkonganti@1: 
kkonganti@1:         DB_PER_SNP_CLUSTER
kkonganti@1:             .out
kkonganti@1:             .asm_chunk_snp_tbl
kkonganti@1:             .concat( DB_PER_COMPUTED_SEROTYPE.out.asm_chunk_comp_tbl )
kkonganti@1:             .map { acc ->
kkonganti@1:                 [ acc.name.find(/\_comp|\_snp/), acc ]
kkonganti@1:             }
kkonganti@1:             .set { ch_index_metadata }
kkonganti@1: 
kkonganti@1:         INDEX_METADATA ( ch_index_metadata )
kkonganti@1: 
kkonganti@1:         DB_PER_COMPUTED_SEROTYPE.out.accs_comp
kkonganti@1:             .concat( DB_PER_SNP_CLUSTER.out.accs_snp )
kkonganti@1:             .collectFile(name: 'accs_to_download.txt')
kkonganti@1:             .splitText(by: params.genomes_chunk, file: true)
kkonganti@1:             .set { ch_accs_to_download }
kkonganti@1:         
kkonganti@1:         SCAFFOLD_GENOMES ( ch_accs_to_download )
kkonganti@1: 
kkonganti@1:         SCAFFOLD_GENOMES
kkonganti@1:             .out
kkonganti@1:             .genomes_dir
kkonganti@1:             .toSortedList()
kkonganti@1:             .flatten()
kkonganti@1:             .unique()
kkonganti@1:             .set { ch_genomes_dir }
kkonganti@1: 
kkonganti@1:         MASH_SKETCH ( 
kkonganti@1:             ch_mash_these_genomes.combine( ch_genomes_dir )
kkonganti@1:         )
kkonganti@1: 
kkonganti@1:         DUMP_SOFTWARE_VERSIONS (
kkonganti@1:             software_versions
kkonganti@1:                 .mix (
kkonganti@1:                     DOWNLOAD_PDG_METADATA.out.versions,
kkonganti@1:                     FILTER_PDG_METADATA.out.versions,
kkonganti@1:                     DB_PER_COMPUTED_SEROTYPE.out.versions,
kkonganti@1:                     DB_PER_SNP_CLUSTER.out.versions,
kkonganti@1:                     INDEX_METADATA.out.versions,
kkonganti@1:                     SCAFFOLD_GENOMES.out.versions,
kkonganti@1:                     MASH_SKETCH.out.versions,
kkonganti@1:                 )
kkonganti@1:                 .unique()
kkonganti@1:                 .collectFile(name: 'collected_versions.yml')
kkonganti@1:         )
kkonganti@1: }
kkonganti@1: 
kkonganti@1: /*
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1:     ON COMPLETE, SHOW GORY DETAILS OF ALL PARAMS WHICH WILL BE HELPFUL TO DEBUG
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1: */
kkonganti@1: 
kkonganti@1: workflow.onComplete {
kkonganti@1:     if (workflow.success) {
kkonganti@1:         sendMail()
kkonganti@1:     }
kkonganti@1: }
kkonganti@1: 
kkonganti@1: workflow.onError {
kkonganti@1:     sendMail()
kkonganti@1: }
kkonganti@1: 
kkonganti@1: /*
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1:     HELP TEXT METHODS FOR BETTERCALLSAL_DB WORKFLOW
kkonganti@1: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@1: */
kkonganti@1: 
kkonganti@1: def help() {
kkonganti@1: 
kkonganti@1:     Map helptext = [:]
kkonganti@1: 
kkonganti@1:     helptext.putAll (
kkonganti@1:         fastqEntryPointHelp().findAll {
kkonganti@1:             it.key =~ /Required|output|Other|Workflow|Author|Version/
kkonganti@1:         } +
kkonganti@1:         wcompHelp(params).text +
kkonganti@1:         wsnpHelp(params).text +
kkonganti@1:         mashsketchHelp(params).text + 
kkonganti@1:         wrapUpHelp()
kkonganti@1:     )
kkonganti@1: 
kkonganti@1:     return addPadding(helptext)
kkonganti@1: }