Mercurial > repos > kkonganti > cfsan_cronology
comparison 0.2.0/workflows/cronology.nf @ 11:a5f31c44f8c9
planemo upload
author | kkonganti |
---|---|
date | Mon, 15 Jul 2024 16:11:44 -0400 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
10:ddf7a172bf30 | 11:a5f31c44f8c9 |
---|---|
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; conciseHelp; \ | |
8 addPadding; wrapUpHelp } from "${params.routines}" | |
9 include { dpubmlstpyHelp } from "${params.toolshelp}${params.fs}dpubmlstpy" | |
10 include { fastpHelp } from "${params.toolshelp}${params.fs}fastp" | |
11 include { mashscreenHelp } from "${params.toolshelp}${params.fs}mashscreen" | |
12 include { tuspyHelp } from "${params.toolshelp}${params.fs}tuspy" | |
13 include { spadesHelp } from "${params.toolshelp}${params.fs}spades" | |
14 include { shovillHelp } from "${params.toolshelp}${params.fs}shovill" | |
15 include { polypolishHelp } from "${params.toolshelp}${params.fs}polypolish" | |
16 include { mashtreeHelp } from "${params.toolshelp}${params.fs}mashtree" | |
17 include { quastHelp } from "${params.toolshelp}${params.fs}quast" | |
18 include { prodigalHelp } from "${params.toolshelp}${params.fs}prodigal" | |
19 include { prokkaHelp } from "${params.toolshelp}${params.fs}prokka" | |
20 include { pirateHelp } from "${params.toolshelp}${params.fs}pirate" | |
21 include { mlstHelp } from "${params.toolshelp}${params.fs}mlst" | |
22 include { abricateHelp } from "${params.toolshelp}${params.fs}abricate" | |
23 | |
24 // Exit if help requested before any subworkflows | |
25 if (params.help) { | |
26 log.info help() | |
27 exit 0 | |
28 } | |
29 | |
30 // Include any necessary modules and subworkflows | |
31 include { PROCESS_FASTQ } from "${params.subworkflows}${params.fs}process_fastq" | |
32 include { PRODKA } from "${params.subworkflows}${params.fs}prodka" | |
33 include { DOWNLOAD_PUBMLST_SCHEME } from "${params.modules}${params.fs}download_pubmlst_scheme${params.fs}main" | |
34 include { DOWNLOAD_REF_GENOME } from "${params.modules}${params.fs}download_ref_genome${params.fs}main" | |
35 include { FASTP } from "${params.modules}${params.fs}fastp${params.fs}main" | |
36 include { MASH_SCREEN } from "${params.modules}${params.fs}mash${params.fs}screen${params.fs}main" | |
37 include { TOP_UNIQUE_SEROVARS } from "${params.modules}${params.fs}top_unique_serovars${params.fs}main" | |
38 include { CAT_UNIQUE } from "${params.modules}${params.fs}cat${params.fs}unique${params.fs}main" | |
39 include { SPADES_ASSEMBLE } from "${params.modules}${params.fs}spades${params.fs}assemble${params.fs}main" | |
40 include { SHOVILL } from "${params.modules}${params.fs}shovill${params.fs}main" | |
41 include { BWA_IDX_MEM } from "${params.modules}${params.fs}custom${params.fs}bwa_idx_mem${params.fs}main" | |
42 include { POLYPOLISH } from "${params.modules}${params.fs}polypolish${params.fs}main" | |
43 include { QUAST } from "${params.modules}${params.fs}quast${params.fs}main" | |
44 include { RMLST_POST } from "${params.modules}${params.fs}rmlst${params.fs}main" | |
45 include { PIRATE } from "${params.modules}${params.fs}pirate${params.fs}main" | |
46 include { MASHTREE } from "${params.modules}${params.fs}mashtree${params.fs}main" | |
47 include { UPLOAD_MICROREACT } from "${params.modules}${params.fs}upload_microreact${params.fs}main" | |
48 include { MLST } from "${params.modules}${params.fs}mlst${params.fs}main" | |
49 include { ABRICATE_RUN } from "${params.modules}${params.fs}abricate${params.fs}run${params.fs}main" | |
50 include { ABRICATE_SUMMARY } from "${params.modules}${params.fs}abricate${params.fs}summary${params.fs}main" | |
51 include { TABLE_SUMMARY } from "${params.modules}${params.fs}cat${params.fs}tables${params.fs}main" | |
52 include { DUMP_SOFTWARE_VERSIONS } from "${params.modules}${params.fs}custom${params.fs}dump_software_versions${params.fs}main" | |
53 include { MULTIQC } from "${params.modules}${params.fs}multiqc${params.fs}main" | |
54 | |
55 /* | |
56 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
57 INPUTS AND ANY CHECKS FOR THE CRONOLOGY WORKFLOW | |
58 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
59 */ | |
60 def spades_custom_hmm = (params.spades_hmm ? file ( "${params.spades_hmm}" ) : false) | |
61 def reads_platform = 0 | |
62 def abricate_dbs = [ 'ncbiamrplus', 'resfinder', 'megares', 'argannot' ] | |
63 | |
64 reads_platform += (params.input ? 1 : 0) | |
65 | |
66 if (params.upload_microreact) { | |
67 checkMetadataExists(params.microreact_api_key, 'Microreact API Key') | |
68 } | |
69 | |
70 if (spades_custom_hmm && !spades_custom_hmm.exists()) { | |
71 stopNow("Please check if the following SPAdes' custom HMM directory\n" + | |
72 "path is valid:\n${params.spades_hmm}\nCannot proceed further!") | |
73 } | |
74 | |
75 if (reads_platform < 1 || reads_platform == 0) { | |
76 stopNow("Please mention at least one absolute path to input folder which contains\n" + | |
77 "FASTQ files sequenced using the --input option.\n" + | |
78 "Ex: --input (Illumina or Generic short reads in FASTQ format)") | |
79 } | |
80 | |
81 /* | |
82 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
83 RUN THE CRONOLOGY WORKFLOW | |
84 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
85 */ | |
86 | |
87 workflow CRONOLOGY { | |
88 main: | |
89 ch_mqc_custom_tbl = Channel.empty() | |
90 ch_dummy = Channel.fromPath("${params.dummyfile}") | |
91 ch_dummy2 = Channel.fromPath("${params.dummyfile2}") | |
92 | |
93 log.info summaryOfParams() | |
94 | |
95 PROCESS_FASTQ() | |
96 | |
97 PROCESS_FASTQ.out.versions | |
98 .set { software_versions } | |
99 | |
100 PROCESS_FASTQ.out.processed_reads | |
101 .map { meta, fastq -> | |
102 meta.sequence_sketch = (params.mash_sketch ?: null) | |
103 [meta, fastq] | |
104 } | |
105 .set { ch_processed_reads } | |
106 | |
107 DOWNLOAD_PUBMLST_SCHEME( params.dpubmlstpy_org ?: null ) | |
108 | |
109 DOWNLOAD_REF_GENOME( | |
110 (params.ref_acc ? ['id': params.ref_acc] : null) | |
111 ) | |
112 | |
113 FASTP( ch_processed_reads ) | |
114 | |
115 FASTP.out.passed_reads | |
116 .set { ch_processed_reads } | |
117 | |
118 FASTP.out.json | |
119 .map { meta, json -> [ json ] } | |
120 .collect() | |
121 .set { ch_multiqc } | |
122 | |
123 MASH_SCREEN( ch_processed_reads ) | |
124 | |
125 TOP_UNIQUE_SEROVARS( MASH_SCREEN.out.screened ) | |
126 | |
127 TOP_UNIQUE_SEROVARS.out.tsv | |
128 .map { meta, tsv -> tsv } | |
129 .collectFile( | |
130 name: 'iTOL_metadata_w_dups.txt', | |
131 keepHeader: true, | |
132 skip: 4, | |
133 sort: true | |
134 ) | |
135 .map { file -> | |
136 def meta = [:] | |
137 meta.id = 'Unique iTOL Metadata' | |
138 meta.skip_header = 4 | |
139 [meta, file] | |
140 } | |
141 .concat( | |
142 TOP_UNIQUE_SEROVARS.out.popup | |
143 .map { meta, popup -> popup } | |
144 .collectFile( | |
145 name: 'iTOL_2_NCBI_Pathogens_w_dups.txt', | |
146 keepHeader: true, | |
147 skip: 3, | |
148 sort: true | |
149 ) | |
150 .map { file -> | |
151 def meta = [:] | |
152 meta.id = 'Unique iTOL Popup' | |
153 meta.skip_header = 3 | |
154 [meta, file] | |
155 } | |
156 ) | |
157 .set { ch_uniq } | |
158 | |
159 TOP_UNIQUE_SEROVARS.out.accessions | |
160 .map { meta, acc -> acc } | |
161 .splitText() | |
162 .collect() | |
163 .flatten() | |
164 .unique() | |
165 .collectFile(name: 'tree_genomes.txt') | |
166 .map { genomes -> [ [id: 'hitsTree'], genomes ]} | |
167 .set { ch_genomes_fofn } | |
168 | |
169 CAT_UNIQUE( ch_uniq ) | |
170 | |
171 if (params.fq_single_end) { | |
172 SPADES_ASSEMBLE( | |
173 ch_processed_reads | |
174 .combine(ch_dummy) | |
175 .combine(ch_dummy2) | |
176 ) | |
177 | |
178 SPADES_ASSEMBLE.out.assembly | |
179 .set{ ch_assembly } | |
180 | |
181 software_versions | |
182 .mix( SPADES_ASSEMBLE.out.versions.ifEmpty(null) ) | |
183 .set { software_versions } | |
184 } else { | |
185 SHOVILL( ch_processed_reads ) | |
186 | |
187 SHOVILL.out.contigs | |
188 .set { ch_assembly } | |
189 | |
190 software_versions | |
191 .mix( SHOVILL.out.versions.ifEmpty(null) ) | |
192 .set { software_versions } | |
193 } | |
194 | |
195 if (params.polypolish_run) { | |
196 BWA_IDX_MEM( | |
197 ch_assembly | |
198 .join( ch_processed_reads ) | |
199 ) | |
200 | |
201 POLYPOLISH( | |
202 ch_assembly | |
203 .join( BWA_IDX_MEM.out.aligned_sam ) | |
204 ) | |
205 | |
206 POLYPOLISH.out.polished | |
207 .set { ch_assembly } | |
208 | |
209 software_versions | |
210 .mix( | |
211 BWA_IDX_MEM.out.versions, | |
212 POLYPOLISH.out.versions | |
213 ) | |
214 .set { software_versions } | |
215 } | |
216 | |
217 ch_assembly | |
218 .combine( DOWNLOAD_REF_GENOME.out.fasta ) | |
219 .combine( DOWNLOAD_REF_GENOME.out.gff ) | |
220 .multiMap { meta, consensus, fasta, gff -> | |
221 sample_fa: consensus | |
222 polished: [meta, consensus] | |
223 ref_fasta: [meta, fasta] | |
224 ref_gff: [meta, gff] | |
225 } | |
226 .set { ch_quast } | |
227 | |
228 MASHTREE( | |
229 ch_genomes_fofn, | |
230 DOWNLOAD_REF_GENOME.out.fasta | |
231 .concat( ch_quast.sample_fa ) | |
232 .collect() | |
233 ) | |
234 | |
235 if (params.upload_microreact) { | |
236 UPLOAD_MICROREACT( | |
237 MASHTREE.out.nwk | |
238 .map { meta, nwk -> | |
239 nwk | |
240 }, | |
241 CAT_UNIQUE.out.csv | |
242 ) | |
243 | |
244 software_versions | |
245 .mix( | |
246 UPLOAD_MICROREACT.out.versions | |
247 ) | |
248 .set { software_versions } | |
249 } | |
250 | |
251 PRODKA( | |
252 ch_quast.ref_fasta, | |
253 ch_quast.polished | |
254 ) | |
255 | |
256 RMLST_POST( ch_assembly ) | |
257 | |
258 MLST ( | |
259 ch_assembly | |
260 .combine( DOWNLOAD_PUBMLST_SCHEME.out.pubmlst_dir ) | |
261 ) | |
262 | |
263 QUAST( | |
264 ch_quast.polished, | |
265 ch_quast.ref_fasta, | |
266 ch_quast.ref_gff | |
267 ) | |
268 | |
269 if (params.pirate_run) { | |
270 PIRATE( | |
271 PRODKA.out.prokka_gff | |
272 .map { meta, gff -> | |
273 tuple( [id: 'Predicted Genes'], gff ) | |
274 } | |
275 .groupTuple(by: [0]) | |
276 ) | |
277 | |
278 software_versions | |
279 .mix( PIRATE.out.versions ) | |
280 .set { software_versions } | |
281 } | |
282 | |
283 RMLST_POST.out.tsv | |
284 .map { meta, tsv -> [ 'rmlst', tsv] } | |
285 .groupTuple(by: [0]) | |
286 .map { it -> tuple ( it[0], it[1].flatten() ) } | |
287 .set { ch_mqc_rmlst_tbl } | |
288 | |
289 MLST.out.tsv | |
290 .map { meta, tsv -> [ 'mlst', tsv] } | |
291 .groupTuple(by: [0]) | |
292 .map { it -> tuple ( it[0], it[1].flatten() ) } | |
293 .set { ch_mqc_custom_tbl } | |
294 | |
295 ABRICATE_RUN ( ch_assembly, abricate_dbs ) | |
296 | |
297 ABRICATE_RUN.out.abricated | |
298 .map { meta, abres -> [ abricate_dbs, abres ] } | |
299 .groupTuple(by: [0]) | |
300 .map { it -> tuple ( it[0], it[1].flatten() ) } | |
301 .set { ch_abricated } | |
302 | |
303 ABRICATE_SUMMARY ( ch_abricated ) | |
304 | |
305 ch_mqc_custom_tbl | |
306 .concat ( | |
307 ch_mqc_rmlst_tbl, | |
308 ABRICATE_SUMMARY.out.ncbiamrplus.map { it -> tuple ( it[0], it[1] )}, | |
309 ABRICATE_SUMMARY.out.resfinder.map { it -> tuple ( it[0], it[1] )}, | |
310 ABRICATE_SUMMARY.out.megares.map { it -> tuple ( it[0], it[1] )}, | |
311 ABRICATE_SUMMARY.out.argannot.map { it -> tuple ( it[0], it[1] )}, | |
312 ) | |
313 .groupTuple(by: [0]) | |
314 .map { it -> [ it[0], it[1].flatten() ]} | |
315 .set { ch_mqc_custom_tbl } | |
316 | |
317 TABLE_SUMMARY ( ch_mqc_custom_tbl ) | |
318 | |
319 DUMP_SOFTWARE_VERSIONS ( | |
320 software_versions | |
321 .mix( | |
322 DOWNLOAD_PUBMLST_SCHEME.out.versions, | |
323 DOWNLOAD_REF_GENOME.out.versions, | |
324 FASTP.out.versions, | |
325 MASH_SCREEN.out.versions, | |
326 TOP_UNIQUE_SEROVARS.out.versions, | |
327 CAT_UNIQUE.out.versions, | |
328 MASHTREE.out.versions, | |
329 POLYPOLISH.out.versions, | |
330 QUAST.out.versions, | |
331 PRODKA.out.versions, | |
332 RMLST_POST.out.versions, | |
333 MLST.out.versions, | |
334 ABRICATE_RUN.out.versions, | |
335 ABRICATE_SUMMARY.out.versions, | |
336 TABLE_SUMMARY.out.versions | |
337 ) | |
338 .unique() | |
339 .collectFile(name: 'collected_versions.yml') | |
340 ) | |
341 | |
342 DUMP_SOFTWARE_VERSIONS.out.mqc_yml | |
343 .concat ( | |
344 ch_multiqc, | |
345 TABLE_SUMMARY.out.mqc_yml, | |
346 PRODKA.out.prokka_txt.map { meta, txt -> txt }, | |
347 QUAST.out.results.map { meta, res -> res } | |
348 ) | |
349 .collect() | |
350 .set { ch_multiqc } | |
351 | |
352 MULTIQC( ch_multiqc ) | |
353 } | |
354 | |
355 /* | |
356 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
357 ON COMPLETE, SHOW GORY DETAILS OF ALL PARAMS WHICH WILL BE HELPFUL TO DEBUG | |
358 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
359 */ | |
360 | |
361 workflow.onComplete { | |
362 if (workflow.success) { | |
363 sendMail() | |
364 } | |
365 } | |
366 | |
367 workflow.onError { | |
368 sendMail() | |
369 } | |
370 | |
371 /* | |
372 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
373 METHOD TO CHECK FILE EXISTENCE | |
374 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
375 */ | |
376 | |
377 def checkMetadataExists(file_path, msg) { | |
378 file_path_obj = file( file_path ) | |
379 | |
380 if (msg.toString().find(/(?i)KMA/)) { | |
381 if (!file_path_obj.parent.exists() || file_path_obj.parent.size() == 0) { | |
382 stopNow("Please check if your ${msg}\n" + | |
383 "[ ${file_path} ]\nexists and that the files are not of size 0.") | |
384 } | |
385 } | |
386 else if (!file_path_obj.exists() || file_path_obj.size() == 0) { | |
387 stopNow("Please check if your ${msg} file\n" + | |
388 "[ ${file_path} ]\nexists and is not of size 0.") | |
389 } | |
390 } | |
391 | |
392 /* | |
393 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
394 HELP TEXT METHODS FOR CRONOLOGY WORKFLOW | |
395 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
396 */ | |
397 | |
398 def help() { | |
399 | |
400 Map helptext = [:] | |
401 Map nH = [:] | |
402 Map fastpAdapterHelp = [:] | |
403 def uHelp = (params.help.getClass().toString() =~ /String/ ? params.help.tokenize(',').join(' ') : '') | |
404 | |
405 Map defaultHelp = [ | |
406 '--help dpubmlstpy' : 'Show dl_pubmlst_profiles_and_schemes.py CLI options', | |
407 '--help fastp' : 'Show fastp CLI options', | |
408 '--help spades' : 'Show spades CLI options', | |
409 '--help shovill' : 'Show shovill CLI options', | |
410 '--help polypolish' : 'Show polypolish CLI options', | |
411 '--help quast' : 'Show quast.py CLI options', | |
412 '--help prodigal' : 'Show prodigal CLI options', | |
413 '--help prokka' : 'Show prokka CLI options', | |
414 '--help pirate' : 'Show priate CLI options', | |
415 '--help mlst' : 'Show mlst CLI options', | |
416 '--help mash' : 'Show mash `screen` CLI options', | |
417 '--help tree' : 'Show mashtree CLI options', | |
418 '--help abricate' : 'Show abricate CLI options\n' | |
419 ] | |
420 | |
421 fastpAdapterHelp['--fastp_use_custom_adapaters'] = "Use custom adapter FASTA with fastp on top of " + | |
422 "built-in adapter sequence auto-detection. Enabling this option will attempt to find and remove " + | |
423 "all possible Illumina adapter and primer sequences but will make the workflow run slow. " + | |
424 "Default: ${params.fastp_use_custom_adapters}" | |
425 | |
426 if (params.help.getClass().toString() =~ /Boolean/ || uHelp.size() == 0) { | |
427 println conciseHelp('fastp,polypolish') | |
428 helptext.putAll(defaultHelp) | |
429 } else { | |
430 params.help.tokenize(',').each { h -> | |
431 if (defaultHelp.keySet().findAll{ it =~ /(?i)\b${h}\b/ }.size() == 0) { | |
432 println conciseHelp('fastp,polypolish') | |
433 stopNow("Tool [ ${h} ] is not a part of ${params.pipeline} pipeline.") | |
434 } | |
435 } | |
436 | |
437 helptext.putAll( | |
438 fastqEntryPointHelp() + | |
439 (uHelp =~ /(?i)\bdpubmlstpy/ ? dpubmlstpyHelp(params).text : nH) + | |
440 (uHelp =~ /(?i)\bfastp/ ? fastpHelp(params).text + fastpAdapterHelp : nH) + | |
441 (uHelp =~ /(?i)\bmash/ ? mashscreenHelp(params).text : nH) + | |
442 (uHelp =~ /(?i)\btuspy/ ? tuspyHelp(params).text : nH) + | |
443 (uHelp =~ /(?i)\bspades/ ? spadesHelp(params).text : nH) + | |
444 (uHelp =~ /(?i)\bshovill/ ? shovillHelp(params).text : nH) + | |
445 (uHelp =~ /(?i)\bpolypolish/ ? polypolishHelp(params).text : nH) + | |
446 (uHelp =~ /(?i)\bquast/ ? quastHelp(params).text : nH) + | |
447 (uHelp =~ /(?i)\bprodigal/ ? prodigalHelp(params).text : nH) + | |
448 (uHelp =~ /(?i)\bprokka/ ? prokkaHelp(params).text : nH) + | |
449 (uHelp =~ /(?i)\bpirate/ ? pirateHelp(params).text : nH) + | |
450 (uHelp =~ /(?i)\bmlst/ ? mlstHelp(params).text : nH) + | |
451 (uHelp =~ /(?i)\btree/ ? mashtreeHelp(params).text : nH) + | |
452 (uHelp =~ /(?i)\babricate/ ? abricateHelp(params).text : nH) + | |
453 wrapUpHelp() | |
454 ) | |
455 } | |
456 | |
457 return addPadding(helptext) | |
458 } |