Mercurial > repos > kkonganti > cfsan_bettercallsal

annotate 0.7.0/workflows/bettercallsal_db.nf @ 20:133571bf891f

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload
author kkonganti
date Mon, 15 Jul 2024 11:42:00 -0400
parents 0e7a0053e4a6
children
rev   line source
kkonganti@17 1 // Define any required imports for this specific workflow
kkonganti@17 2 import java.nio.file.Paths
kkonganti@17 3 import nextflow.file.FileHelper
kkonganti@17 4
kkonganti@17 5
kkonganti@17 6 // Include any necessary methods
kkonganti@17 7 include { \
kkonganti@17 8 fastqEntryPointHelp; summaryOfParams; stopNow; sendMail; \
kkonganti@17 9 addPadding; wrapUpHelp } from "${params.routines}"
kkonganti@17 10 include { wcompHelp } from "${params.toolshelp}${params.fs}wcomp"
kkonganti@17 11 include { wsnpHelp } from "${params.toolshelp}${params.fs}wsnp"
kkonganti@17 12 include { mashsketchHelp } from "${params.toolshelp}${params.fs}mashsketch"
kkonganti@17 13
kkonganti@17 14
kkonganti@17 15 // Exit if help requested before any subworkflows
kkonganti@17 16 if (params.help) {
kkonganti@17 17 log.info help()
kkonganti@17 18 exit 0
kkonganti@17 19 }
kkonganti@17 20
kkonganti@17 21
kkonganti@17 22 // Include any necessary modules and subworkflows
kkonganti@17 23 include { DOWNLOAD_PDG_METADATA } from "${params.modules}${params.fs}download_pdg_metadata${params.fs}main"
kkonganti@17 24 include { FILTER_PDG_METADATA } from "${params.modules}${params.fs}filter_pdg_metadata${params.fs}main"
kkonganti@17 25 include { DB_PER_COMPUTED_SEROTYPE } from "${params.modules}${params.fs}db_per_computed_serotype${params.fs}main"
kkonganti@17 26 include { DB_PER_SNP_CLUSTER } from "${params.modules}${params.fs}db_per_snp_cluster${params.fs}main"
kkonganti@17 27 include { INDEX_METADATA } from "${params.modules}${params.fs}index_metadata${params.fs}main"
kkonganti@17 28 include { SCAFFOLD_GENOMES } from "${params.modules}${params.fs}scaffold_genomes${params.fs}main"
kkonganti@17 29 include { MASH_SKETCH } from "${params.modules}${params.fs}mash${params.fs}sketch${params.fs}main"
kkonganti@17 30 include { DUMP_SOFTWARE_VERSIONS } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main"
kkonganti@17 31
kkonganti@17 32 /*
kkonganti@17 33 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 34 INPUTS AND ANY CHECKS FOR THE BETTERCALLSAL_DB WORKFLOW
kkonganti@17 35 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 36 */
kkonganti@17 37
kkonganti@17 38 if (!params.output) {
kkonganti@17 39 stopNow("Please mention the absolute UNIX path to store the DB flat files\n" +
kkonganti@17 40 "using the --output option.\n" +
kkonganti@17 41 "Ex: --output /path/to/bettercallsal/db_files")
kkonganti@17 42 }
kkonganti@17 43
kkonganti@17 44 /*
kkonganti@17 45 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 46 RUN THE BETTERCALLSAL_DB WORKFLOW
kkonganti@17 47 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 48 */
kkonganti@17 49
kkonganti@17 50 workflow BETTERCALLSAL_DB {
kkonganti@17 51 main:
kkonganti@17 52 log.info summaryOfParams()
kkonganti@17 53
kkonganti@17 54 DOWNLOAD_PDG_METADATA ( params.pdg_release ?: null )
kkonganti@17 55
kkonganti@17 56 DOWNLOAD_PDG_METADATA
kkonganti@17 57 .out
kkonganti@17 58 .versions
kkonganti@17 59 .set { software_versions }
kkonganti@17 60
kkonganti@17 61 FILTER_PDG_METADATA (
kkonganti@17 62 DOWNLOAD_PDG_METADATA.out.accs
kkonganti@17 63 .splitText(by: params.genomes_chunk.toInteger() * 10, file: true)
kkonganti@17 64 )
kkonganti@17 65
kkonganti@17 66 DB_PER_COMPUTED_SEROTYPE (
kkonganti@17 67 FILTER_PDG_METADATA.out.accs_chunk_tbl
kkonganti@17 68 .collectFile(name: 'per_comp_db_accs.txt'),
kkonganti@17 69 DOWNLOAD_PDG_METADATA.out.pdg_metadata
kkonganti@17 70 )
kkonganti@17 71
kkonganti@17 72 DB_PER_SNP_CLUSTER (
kkonganti@17 73 FILTER_PDG_METADATA.out.accs_chunk_tbl
kkonganti@17 74 .collectFile(name: 'per_snp_db_accs.txt'),
kkonganti@17 75 DOWNLOAD_PDG_METADATA.out.pdg_metadata,
kkonganti@17 76 DOWNLOAD_PDG_METADATA.out.snp_cluster_metadata
kkonganti@17 77 )
kkonganti@17 78
kkonganti@17 79 DB_PER_COMPUTED_SEROTYPE.out.genome_paths
kkonganti@17 80 .map { query ->
kkonganti@17 81 kv = [:]
kkonganti@17 82 kv['id'] = 'comp'
kkonganti@17 83 [ kv, query ]
kkonganti@17 84 }
kkonganti@17 85 .concat(
kkonganti@17 86 DB_PER_SNP_CLUSTER.out.genome_paths
kkonganti@17 87 .map { query ->
kkonganti@17 88 kv = [:]
kkonganti@17 89 kv['id'] = 'snp'
kkonganti@17 90 [ kv, query ]
kkonganti@17 91 }
kkonganti@17 92 )
kkonganti@17 93 .groupTuple(by: [0])
kkonganti@17 94 .set { ch_mash_these_genomes }
kkonganti@17 95
kkonganti@17 96 DB_PER_SNP_CLUSTER
kkonganti@17 97 .out
kkonganti@17 98 .asm_chunk_snp_tbl
kkonganti@17 99 .concat( DB_PER_COMPUTED_SEROTYPE.out.asm_chunk_comp_tbl )
kkonganti@17 100 .map { acc -> [ acc.name.find(/\_comp|\_snp/), acc ] }
kkonganti@17 101 .set { ch_index_metadata }
kkonganti@17 102
kkonganti@17 103 INDEX_METADATA ( ch_index_metadata )
kkonganti@17 104
kkonganti@17 105 DB_PER_COMPUTED_SEROTYPE.out.accs_comp
kkonganti@17 106 .concat( DB_PER_SNP_CLUSTER.out.accs_snp )
kkonganti@17 107 .splitText()
kkonganti@17 108 .collect()
kkonganti@17 109 .flatten()
kkonganti@17 110 .unique()
kkonganti@17 111 .collectFile(name: 'accs_to_download.txt')
kkonganti@17 112 .splitText(by: params.genomes_chunk, file: true)
kkonganti@17 113 .set { ch_accs_to_download }
kkonganti@17 114
kkonganti@17 115 SCAFFOLD_GENOMES ( ch_accs_to_download )
kkonganti@17 116
kkonganti@17 117 SCAFFOLD_GENOMES
kkonganti@17 118 .out
kkonganti@17 119 .genomes_dir
kkonganti@17 120 .toSortedList()
kkonganti@17 121 .flatten()
kkonganti@17 122 .unique()
kkonganti@17 123 .set { ch_genomes_dir }
kkonganti@17 124
kkonganti@17 125 MASH_SKETCH (
kkonganti@17 126 ch_mash_these_genomes.combine( ch_genomes_dir )
kkonganti@17 127 )
kkonganti@17 128
kkonganti@17 129 DUMP_SOFTWARE_VERSIONS (
kkonganti@17 130 software_versions
kkonganti@17 131 .mix (
kkonganti@17 132 DOWNLOAD_PDG_METADATA.out.versions,
kkonganti@17 133 FILTER_PDG_METADATA.out.versions,
kkonganti@17 134 DB_PER_COMPUTED_SEROTYPE.out.versions,
kkonganti@17 135 DB_PER_SNP_CLUSTER.out.versions,
kkonganti@17 136 INDEX_METADATA.out.versions,
kkonganti@17 137 SCAFFOLD_GENOMES.out.versions,
kkonganti@17 138 MASH_SKETCH.out.versions,
kkonganti@17 139 )
kkonganti@17 140 .unique()
kkonganti@17 141 .collectFile(name: 'collected_versions.yml')
kkonganti@17 142 )
kkonganti@17 143 }
kkonganti@17 144
kkonganti@17 145 /*
kkonganti@17 146 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 147 ON COMPLETE, SHOW GORY DETAILS OF ALL PARAMS WHICH WILL BE HELPFUL TO DEBUG
kkonganti@17 148 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 149 */
kkonganti@17 150
kkonganti@17 151 workflow.onComplete {
kkonganti@17 152 if (workflow.success) {
kkonganti@17 153 sendMail()
kkonganti@17 154 }
kkonganti@17 155 }
kkonganti@17 156
kkonganti@17 157 workflow.onError {
kkonganti@17 158 sendMail()
kkonganti@17 159 }
kkonganti@17 160
kkonganti@17 161 /*
kkonganti@17 162 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 163 HELP TEXT METHODS FOR BETTERCALLSAL_DB WORKFLOW
kkonganti@17 164 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
kkonganti@17 165 */
kkonganti@17 166
kkonganti@17 167 def help() {
kkonganti@17 168
kkonganti@17 169 Map helptext = [:]
kkonganti@17 170
kkonganti@17 171 helptext.putAll (
kkonganti@17 172 fastqEntryPointHelp().findAll {
kkonganti@17 173 it.key =~ /Required|output|Other|Workflow|Author|Version/
kkonganti@17 174 } +
kkonganti@17 175 wcompHelp(params).text +
kkonganti@17 176 wsnpHelp(params).text +
kkonganti@17 177 mashsketchHelp(params).text +
kkonganti@17 178 wrapUpHelp()
kkonganti@17 179 )
kkonganti@17 180
kkonganti@17 181 return addPadding(helptext)
kkonganti@17 182 }