Mercurial > repos > kkonganti > cfsan_centriflaken

comparison 0.3.0/workflows/centriflaken_hy.nf @ 92:295c2597a475

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload"
author kkonganti
date Tue, 19 Jul 2022 10:07:24 -0400
parents
children
comparison
equal deleted inserted replaced
91:046e57368d3b 92:295c2597a475
1 // Define any required imports for this specific workflow
2 import java.nio.file.Paths
3 import nextflow.file.FileHelper
4
5 // Include any necessary methods
6 include { \
7 summaryOfParams; stopNow; fastqEntryPointHelp; sendMail; \
8 addPadding; wrapUpHelp } from "${params.routines}"
9 include { kraken2Help } from "${params.toolshelp}${params.fs}kraken2"
10 include { centrifugeHelp } from "${params.toolshelp}${params.fs}centrifuge"
11 include { megahitHelp } from "${params.toolshelp}${params.fs}megahit"
12 include { spadesHelp } from "${params.toolshelp}${params.fs}spades"
13 include { serotypefinderHelp } from "${params.toolshelp}${params.fs}serotypefinder"
14 include { seqsero2Help } from "${params.toolshelp}${params.fs}seqsero2"
15 include { mlstHelp } from "${params.toolshelp}${params.fs}mlst"
16 include { abricateHelp } from "${params.toolshelp}${params.fs}abricate"
17
18 // Exit if help requested before any subworkflows
19 if (params.help) {
20 log.info help()
21 exit 0
22 }
23
24 // Include any necessary modules and subworkflows
25 include { PROCESS_FASTQ } from "${params.subworkflows}${params.fs}process_fastq"
26 include { FASTQC } from "${params.modules}${params.fs}fastqc${params.fs}main"
27 include { CENTRIFUGE_CLASSIFY } from "${params.modules}${params.fs}centrifuge${params.fs}classify${params.fs}main"
28 include { CENTRIFUGE_PROCESS } from "${params.modules}${params.fs}centrifuge${params.fs}process${params.fs}main"
29 include { SEQKIT_GREP } from "${params.modules}${params.fs}seqkit${params.fs}grep${params.fs}main"
30 include { MEGAHIT_ASSEMBLE } from "${params.modules}${params.fs}megahit${params.fs}assemble${params.fs}main"
31 include { SPADES_ASSEMBLE } from "${params.modules}${params.fs}spades${params.fs}assemble${params.fs}main"
32 include { KRAKEN2_CLASSIFY } from "${params.modules}${params.fs}kraken2${params.fs}classify${params.fs}main"
33 include { KRAKEN2_EXTRACT_CONTIGS } from "${params.modules}${params.fs}kraken2${params.fs}extract_contigs${params.fs}main"
34 include { SEROTYPEFINDER } from "${params.modules}${params.fs}serotypefinder${params.fs}main"
35 include { SEQSERO2 } from "${params.modules}${params.fs}seqsero2${params.fs}main"
36 include { MLST } from "${params.modules}${params.fs}mlst${params.fs}main"
37 include { ABRICATE_RUN } from "${params.modules}${params.fs}abricate${params.fs}run${params.fs}main"
38 include { ABRICATE_SUMMARY } from "${params.modules}${params.fs}abricate${params.fs}summary${params.fs}main"
39 include { TABLE_SUMMARY } from "${params.modules}${params.fs}cat${params.fs}tables${params.fs}main"
40 include { MULTIQC } from "${params.modules}${params.fs}multiqc${params.fs}main"
41 include { DUMP_SOFTWARE_VERSIONS } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main"
42
43
44
45 /*
46 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
47 INPUTS AND ANY CHECKS FOR THE CENTRIFLAKEN-HY WORKFLOW
48 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
49 */
50
51 def kraken2_db_dir = file ( "${params.kraken2_db}" )
52 def centrifuge_x = file ( "${params.centrifuge_x}" )
53 def spades_custom_hmm = (params.spades_hmm ? file ( "${params.spades_hmm}" ) : false)
54 def reads_platform = 0
55 def abricate_dbs = [ 'ncbiamrplus', 'resfinder', 'megares', 'argannot' ]
56
57 reads_platform += (params.input ? 1 : 0)
58
59 if (!kraken2_db_dir.exists() || !centrifuge_x.getParent().exists()) {
60 stopNow("Please check if the following absolute paths are valid:\n" +
61 "${params.kraken2_db}\n${params.centrifuge_x}\n" +
62 "Cannot proceed further!")
63 }
64
65 if (spades_custom_hmm && !spades_custom_hmm.exists()) {
66 stopNow("Please check if the following SPAdes' custom HMM directory\n" +
67 "path is valid:\n${params.spades_hmm}\nCannot proceed further!")
68 }
69
70 if (reads_platform < 1 || reads_platform == 0) {
71 stopNow("Please mention at least one absolute path to input folder which contains\n" +
72 "FASTQ files sequenced using the --input option.\n" +
73 "Ex: --input (Illumina or Generic short reads in FASTQ format)")
74 }
75
76 if (params.centrifuge_extract_bug != params.kraken2_extract_bug) {
77 stopNow("Please make sure that the bug to be extracted is same\n" +
78 "for both --centrifuge_extract_bug and --kraken2_extract_bug options.")
79 }
80
81 /*
82 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
83 RUN THE CENTRIFLAKEN-HY WORKFLOW
84 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
85 */
86
87 workflow CENTRIFLAKEN_HY {
88 main:
89 ch_asm_filtered_contigs = Channel.empty()
90 ch_mqc_custom_tbl = Channel.empty()
91 ch_dummy = Channel.fromPath("${params.dummyfile}")
92 ch_dummy2 = Channel.fromPath("${params.dummyfile2}")
93
94 log.info summaryOfParams()
95
96 PROCESS_FASTQ()
97 .processed_reads
98 .map {
99 meta, fastq ->
100 meta.centrifuge_x = params.centrifuge_x
101 meta.kraken2_db = params.kraken2_db
102 [meta, fastq]
103 }
104 .set { ch_processed_reads }
105
106 PROCESS_FASTQ
107 .out
108 .versions
109 .set { software_versions }
110
111 FASTQC ( ch_processed_reads )
112
113 CENTRIFUGE_CLASSIFY ( ch_processed_reads )
114
115 CENTRIFUGE_PROCESS (
116 CENTRIFUGE_CLASSIFY.out.report
117 .join( CENTRIFUGE_CLASSIFY.out.output )
118 )
119
120 ch_processed_reads.join ( CENTRIFUGE_PROCESS.out.extracted )
121 .set { ch_centrifuge_extracted }
122
123 SEQKIT_GREP ( ch_centrifuge_extracted )
124
125 // As of 06/02/2022, with the upcoming newer versions of NextFlow, we will be able to do
126 // allowNull: true for both input and output, but until then, we have to use dummy files.
127 // and work arounds.
128 // https://github.com/nextflow-io/nextflow/pull/2893
129 if (params.spades_run) {
130 SPADES_ASSEMBLE (
131 SEQKIT_GREP.out.fastx
132 .combine(ch_dummy)
133 .combine(ch_dummy2)
134 )
135
136 SPADES_ASSEMBLE
137 .out
138 .assembly
139 .set { ch_assembly }
140
141 software_versions
142 .mix ( SPADES_ASSEMBLE.out.versions.ifEmpty(null) )
143 .set { software_versions }
144 } else if (params.megahit_run) {
145 MEGAHIT_ASSEMBLE (
146 SEQKIT_GREP.out.fastx
147 )
148
149 MEGAHIT_ASSEMBLE
150 .out
151 .assembly
152 .set { ch_assembly }
153
154 software_versions
155 .mix ( MEGAHIT_ASSEMBLE.out.versions.ifEmpty(null) )
156 .set { software_versions }
157 }
158
159 ch_assembly
160 .map {
161 meta, fastq ->
162 meta.is_assembly = true
163 [meta, fastq]
164 }
165 .set { ch_assembly }
166
167 ch_assembly.ifEmpty { [ false, false ] }
168
169 KRAKEN2_CLASSIFY ( ch_assembly )
170
171 KRAKEN2_EXTRACT_CONTIGS (
172 ch_assembly
173 .join( KRAKEN2_CLASSIFY.out.kraken_output ),
174 params.kraken2_extract_bug
175 )
176
177 KRAKEN2_EXTRACT_CONTIGS
178 .out
179 .asm_filtered_contigs
180 .map {
181 meta, fastq ->
182 meta.organism = params.kraken2_extract_bug.split(/\s+/)[0].capitalize()
183 meta.serotypefinder_db = params.serotypefinder_db
184 [meta, fastq]
185 }
186 .set { ch_asm_filtered_contigs }
187
188 SEROTYPEFINDER ( ch_asm_filtered_contigs )
189
190 SEQSERO2 ( ch_asm_filtered_contigs )
191
192 MLST ( ch_asm_filtered_contigs )
193
194 ABRICATE_RUN (
195 ch_asm_filtered_contigs,
196 abricate_dbs
197 )
198
199 ABRICATE_RUN
200 .out
201 .abricated
202 .map { meta, abres -> [ abricate_dbs, abres ] }
203 .groupTuple(by: [0])
204 .map { it -> tuple ( it[0], it[1].flatten() ) }
205 .set { ch_abricated }
206
207 ABRICATE_SUMMARY ( ch_abricated )
208
209 CENTRIFUGE_CLASSIFY.out.kreport
210 .map { meta, kreport -> [ kreport ] }
211 .flatten()
212 .concat (
213 KRAKEN2_CLASSIFY.out.kraken_report
214 .map { meta, kreport -> [ kreport ] }
215 .flatten(),
216 FASTQC.out.zip
217 .map { meta, zip -> [ zip ] }
218 .flatten()
219 )
220 .set { ch_mqc_classify }
221
222 if (params.serotypefinder_run) {
223 SEROTYPEFINDER
224 .out
225 .serotyped
226 .map { meta, tsv -> [ 'serotypefinder', tsv ] }
227 .groupTuple(by: [0])
228 .map { it -> tuple ( it[0], it[1].flatten() ) }
229 .set { ch_mqc_custom_tbl }
230 } else if (params.seqsero2_run) {
231 SEQSERO2
232 .out
233 .serotyped
234 .map { meta, tsv -> [ 'seqsero2', tsv ] }
235 .groupTuple(by: [0])
236 .map { it -> tuple ( it[0], it[1].flatten() ) }
237 .set { ch_mqc_custom_tbl }
238 }
239
240 ch_mqc_custom_tbl
241 .concat (
242 ABRICATE_SUMMARY.out.ncbiamrplus.map{ it -> tuple ( it[0], it[1] )},
243 ABRICATE_SUMMARY.out.resfinder.map{ it -> tuple ( it[0], it[1] )},
244 ABRICATE_SUMMARY.out.megares.map{ it -> tuple ( it[0], it[1] )},
245 ABRICATE_SUMMARY.out.argannot.map{ it -> tuple ( it[0], it[1] )},
246 )
247 .groupTuple(by: [0])
248 .map { it -> [ it[0], it[1].flatten() ]}
249 .set { ch_mqc_custom_tbl }
250
251 TABLE_SUMMARY ( ch_mqc_custom_tbl )
252
253 DUMP_SOFTWARE_VERSIONS (
254 software_versions
255 .mix (
256 FASTQC.out.versions,
257 CENTRIFUGE_CLASSIFY.out.versions,
258 CENTRIFUGE_PROCESS.out.versions,
259 SEQKIT_GREP.out.versions,
260 KRAKEN2_CLASSIFY.out.versions.ifEmpty(null),
261 KRAKEN2_EXTRACT_CONTIGS.out.versions.ifEmpty(null),
262 SEROTYPEFINDER.out.versions.ifEmpty(null),
263 SEQSERO2.out.versions.ifEmpty(null),
264 MLST.out.versions.ifEmpty(null),
265 ABRICATE_RUN.out.versions.ifEmpty(null),
266 ABRICATE_SUMMARY.out.versions.ifEmpty(null),
267 TABLE_SUMMARY.out.versions.ifEmpty(null)
268 )
269 .unique()
270 .collectFile(name: 'collected_versions.yml')
271 )
272
273 DUMP_SOFTWARE_VERSIONS
274 .out
275 .mqc_yml
276 .concat (
277 ch_mqc_classify,
278 TABLE_SUMMARY.out.mqc_yml
279 )
280 .collect()
281 .set { ch_multiqc }
282
283 MULTIQC ( ch_multiqc )
284 }
285
286 /*
287 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
288 ON COMPLETE, SHOW GORY DETAILS OF ALL PARAMS WHICH WILL BE HELPFUL TO DEBUG
289 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
290 */
291
292 workflow.onComplete {
293 if (workflow.success) {
294 // CREATE APPROPRIATE DIRECTORIES AND MOVE AS REQUESTED BY STAKEHOLDER(S)
295 //
296 // Nextflow's .moveTo will error out if directories contain files and it
297 // would be complex to include logic to skip directories
298 //
299 def final_intermediate_dir = "${params.output}${params.fs}${params.pipeline}-steps"
300 def final_results_dir = "${params.output}${params.fs}${params.pipeline}-results"
301 def kraken2_ext_contigs = file( "${final_intermediate_dir}${params.fs}kraken2_extract_contigs", type: 'dir' )
302 def final_intermediate = file( final_intermediate_dir, type: 'dir' )
303 def final_results = file( final_results_dir, type: 'dir' )
304 def pipeline_output = file( params.output, type: 'dir' )
305
306 if ( !final_intermediate.exists() ) {
307 final_intermediate.mkdirs()
308
309 FileHelper.visitFiles(Paths.get("${params.output}"), '*') {
310 if ( !(it.name ==~ /^(${params.cfsanpipename}|multiqc|\.nextflow|${workflow.workDir.name}|${params.pipeline}).*/) ) {
311 FileHelper.movePath(
312 it, Paths.get( "${final_intermediate_dir}${params.fs}${it.name}" )
313 )
314 }
315 }
316 }
317
318 if ( kraken2_ext_contigs.exists() && !final_results.exists() ) {
319 final_results.mkdirs()
320
321 FileHelper.movePath(
322 Paths.get( "${final_intermediate_dir}${params.fs}kraken2_extract_contigs" ),
323 Paths.get( "${final_results_dir}${params.fs}kraken2_extract_contigs" )
324 )
325 }
326
327 sendMail()
328 }
329 }
330
331 workflow.onError {
332 sendMail()
333 }
334
335 /*
336 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
337 HELPER METHODS FOR CENTRIFLAKEN-HY WORKFLOW
338 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
339 */
340
341 def help() {
342
343 Map helptext = [:]
344
345 helptext.putAll (
346 fastqEntryPointHelp() +
347 kraken2Help(params).text +
348 centrifugeHelp(params).text +
349 megahitHelp(params).text +
350 spadesHelp(params).text +
351 serotypefinderHelp(params).text +
352 seqsero2Help(params).text +
353 mlstHelp(params).text +
354 abricateHelp(params).text +
355 wrapUpHelp()
356 )
357
358 return addPadding(helptext)
359 }