@@ -321,25 +321,35 @@ process uk_remove_duplicates_rootbiosample_by_date {
321
321
}
322
322
323
323
324
- workflow deduplicate_cog_uk {
324
+ workflow deduplicate_by_cogid_cog_uk {
325
325
take :
326
326
uk_fasta
327
327
uk_metadata
328
328
main :
329
329
uk_annotate_with_unmapped_genome_completeness(uk_fasta, uk_metadata)
330
330
uk_remove_duplicates_COGID_by_proportionN(uk_fasta, uk_annotate_with_unmapped_genome_completeness. out)
331
331
uk_unify_headers(uk_remove_duplicates_COGID_by_proportionN. out. uk_fasta_updated, uk_remove_duplicates_COGID_by_proportionN. out. uk_metadata_updated)
332
- uk_remove_duplicates_biosamplesourceid_by_date(uk_unify_headers. out, uk_remove_duplicates_COGID_by_proportionN. out. uk_metadata_updated)
332
+ emit :
333
+ fasta = uk_unify_headers. out
334
+ metadata = uk_remove_duplicates_COGID_by_proportionN. out. uk_metadata_updated
335
+ }
336
+
337
+ workflow deduplicate_by_biosample_cog_uk {
338
+ take :
339
+ uk_fasta
340
+ uk_metadata
341
+ main :
342
+ uk_remove_duplicates_biosamplesourceid_by_date(uk_fasta, uk_metadata)
333
343
uk_remove_duplicates_rootbiosample_by_date(uk_remove_duplicates_biosamplesourceid_by_date. out. uk_fasta_updated, uk_remove_duplicates_biosamplesourceid_by_date. out. uk_metadata_updated)
334
344
emit :
335
345
fasta = uk_remove_duplicates_rootbiosample_by_date. out. uk_fasta_updated
336
346
metadata = uk_remove_duplicates_rootbiosample_by_date. out. uk_metadata_updated
337
- all_fasta = uk_unify_headers. out
338
347
}
339
348
340
349
341
350
workflow {
342
351
uk_fasta = file(params. uk_fasta)
343
352
uk_metadata = file(params. uk_metadata)
344
- deduplicate_cog_uk(uk_fasta, uk_metadata)
353
+ deduplicate_by_cogid_cog_uk(uk_fasta, uk_metadata)
354
+ deduplicate_by_biosample_cog_uk(deduplicate_by_cogid_cog_uk. out. fasta, deduplicate_by_cogid_cog_uk. out. metadata)
345
355
}
0 commit comments