annotate 1.0.0/workflows/bettercallsal_db.nf @ 0:0a8dda29956e draft default tip

planemo upload
author galaxytrakr
date Thu, 28 May 2026 20:41:10 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
1 // Define any required imports for this specific workflow
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
2 import java.nio.file.Paths
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
3 import nextflow.file.FileHelper
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
4
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
5
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
6 // Include any necessary methods
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
7 include { \
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
8 fastqEntryPointHelp; summaryOfParams; stopNow; sendMail; \
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
9 addPadding; wrapUpHelp } from "${params.routines}"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
10 include { wcompHelp } from "${params.toolshelp}${params.fs}wcomp"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
11 include { wsnpHelp } from "${params.toolshelp}${params.fs}wsnp"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
12 include { mashsketchHelp } from "${params.toolshelp}${params.fs}mashsketch"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
13
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
14
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
15 // Exit if help requested before any subworkflows
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
16 if (params.help) {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
17 log.info help()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
18 exit 0
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
19 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
20
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
21
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
22 // Include any necessary modules and subworkflows
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
23 include { DOWNLOAD_PDG_METADATA } from "${params.modules}${params.fs}download_pdg_metadata${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
24 include { FILTER_PDG_METADATA } from "${params.modules}${params.fs}filter_pdg_metadata${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
25 include { DB_PER_COMPUTED_SEROTYPE } from "${params.modules}${params.fs}db_per_computed_serotype${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
26 include { DB_PER_SNP_CLUSTER } from "${params.modules}${params.fs}db_per_snp_cluster${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
27 include { INDEX_METADATA } from "${params.modules}${params.fs}index_metadata${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
28 include { SCAFFOLD_GENOMES } from "${params.modules}${params.fs}scaffold_genomes${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
29 include { MASH_SKETCH } from "${params.modules}${params.fs}mash${params.fs}sketch${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
30 include { DUMP_SOFTWARE_VERSIONS } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main"
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
31
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
32 /*
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
33 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
34 INPUTS AND ANY CHECKS FOR THE BETTERCALLSAL_DB WORKFLOW
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
35 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
36 */
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
37
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
38 if (!params.output) {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
39 stopNow("Please mention the absolute UNIX path to store the DB flat files\n" +
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
40 "using the --output option.\n" +
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
41 "Ex: --output /path/to/bettercallsal/db_files")
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
42 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
43
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
44 /*
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
45 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
46 RUN THE BETTERCALLSAL_DB WORKFLOW
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
47 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
48 */
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
49
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
50 workflow BETTERCALLSAL_DB {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
51 main:
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
52 log.info summaryOfParams()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
53
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
54 DOWNLOAD_PDG_METADATA ( params.pdg_release ?: null )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
55
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
56 DOWNLOAD_PDG_METADATA
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
57 .out
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
58 .versions
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
59 .set { software_versions }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
60
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
61 FILTER_PDG_METADATA (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
62 DOWNLOAD_PDG_METADATA.out.accs
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
63 .splitText(by: params.genomes_chunk.toInteger() * 10, file: true)
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
64 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
65
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
66 DB_PER_COMPUTED_SEROTYPE (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
67 FILTER_PDG_METADATA.out.accs_chunk_tbl
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
68 .collectFile(name: 'per_comp_db_accs.txt'),
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
69 DOWNLOAD_PDG_METADATA.out.pdg_metadata
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
70 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
71
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
72 DB_PER_SNP_CLUSTER (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
73 FILTER_PDG_METADATA.out.accs_chunk_tbl
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
74 .collectFile(name: 'per_snp_db_accs.txt'),
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
75 DOWNLOAD_PDG_METADATA.out.pdg_metadata,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
76 DOWNLOAD_PDG_METADATA.out.snp_cluster_metadata
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
77 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
78
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
79 DB_PER_COMPUTED_SEROTYPE.out.genome_paths
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
80 .map { query ->
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
81 kv = [:]
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
82 kv['id'] = 'comp'
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
83 [ kv, query ]
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
84 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
85 .concat(
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
86 DB_PER_SNP_CLUSTER.out.genome_paths
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
87 .map { query ->
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
88 kv = [:]
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
89 kv['id'] = 'snp'
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
90 [ kv, query ]
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
91 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
92 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
93 .groupTuple(by: [0])
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
94 .set { ch_mash_these_genomes }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
95
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
96 DB_PER_SNP_CLUSTER
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
97 .out
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
98 .asm_chunk_snp_tbl
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
99 .concat( DB_PER_COMPUTED_SEROTYPE.out.asm_chunk_comp_tbl )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
100 .map { acc -> [ acc.name.find(/\_comp|\_snp/), acc ] }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
101 .set { ch_index_metadata }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
102
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
103 INDEX_METADATA ( ch_index_metadata )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
104
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
105 DB_PER_COMPUTED_SEROTYPE.out.accs_comp
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
106 .concat( DB_PER_SNP_CLUSTER.out.accs_snp )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
107 .splitText()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
108 .collect()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
109 .flatten()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
110 .unique()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
111 .collectFile(name: 'accs_to_download.txt')
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
112 .splitText(by: params.genomes_chunk, file: true)
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
113 .set { ch_accs_to_download }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
114
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
115 SCAFFOLD_GENOMES ( ch_accs_to_download )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
116
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
117 SCAFFOLD_GENOMES
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
118 .out
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
119 .genomes_dir
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
120 .toSortedList()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
121 .flatten()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
122 .unique()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
123 .set { ch_genomes_dir }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
124
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
125 MASH_SKETCH (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
126 ch_mash_these_genomes.combine( ch_genomes_dir )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
127 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
128
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
129 DUMP_SOFTWARE_VERSIONS (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
130 software_versions
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
131 .mix (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
132 DOWNLOAD_PDG_METADATA.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
133 FILTER_PDG_METADATA.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
134 DB_PER_COMPUTED_SEROTYPE.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
135 DB_PER_SNP_CLUSTER.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
136 INDEX_METADATA.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
137 SCAFFOLD_GENOMES.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
138 MASH_SKETCH.out.versions,
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
139 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
140 .unique()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
141 .collectFile(name: 'collected_versions.yml')
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
142 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
143 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
144
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
145 /*
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
146 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
147 ON COMPLETE, SHOW GORY DETAILS OF ALL PARAMS WHICH WILL BE HELPFUL TO DEBUG
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
148 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
149 */
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
150
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
151 workflow.onComplete {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
152 if (workflow.success) {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
153 sendMail()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
154 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
155 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
156
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
157 workflow.onError {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
158 sendMail()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
159 }
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
160
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
161 /*
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
162 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
163 HELP TEXT METHODS FOR BETTERCALLSAL_DB WORKFLOW
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
164 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
165 */
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
166
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
167 def help() {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
168
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
169 Map helptext = [:]
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
170
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
171 helptext.putAll (
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
172 fastqEntryPointHelp().findAll {
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
173 it.key =~ /Required|output|Other|Workflow|Author|Version/
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
174 } +
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
175 wcompHelp(params).text +
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
176 wsnpHelp(params).text +
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
177 mashsketchHelp(params).text +
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
178 wrapUpHelp()
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
179 )
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
180
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
181 return addPadding(helptext)
0a8dda29956e planemo upload
galaxytrakr
parents:
diff changeset
182 }