fixes

Author: Rachel Colquhoun

bin/type_aas_and_dels.py

@@ -14,6 +14,7 @@ def parse_args():
     parser.add_argument('--reference-fasta', dest = 'reference_fasta', required=True, help='Reference FASTA')
     parser.add_argument('--aas', dest = 'aas', required=False, help='CSV of AAs')
     parser.add_argument('--dels', dest = 'dels', required=False, help='CSV of deletions')
+    paraser.add_argument('--index-column', dest = 'index_column', required=False, default='sequence_name')
 
     args = parser.parse_args()
     return args
@@ -65,7 +66,7 @@ def parse_del_file(file, ref_fasta):
 
     return(ls)
 
-def type_aas_and_dels(in_fasta, in_aa_file, in_del_file, reference_fasta, in_metadata, out_metadata):
+def type_aas_and_dels(in_fasta, in_aa_file, in_del_file, reference_fasta, in_metadata, out_metadata, index_column):
     alignment = SeqIO.index(in_fasta, "fasta")
     AAs = parse_AA_file(in_aa_file)
     dels = parse_del_file(in_del_file, reference_fasta)
@@ -81,7 +82,7 @@ def type_aas_and_dels(in_fasta, in_aa_file, in_del_file, reference_fasta, in_met
         writer.writeheader()
 
         for row in reader:
-            id = row["fasta_header"]
+            id = row[index_column]
             seq = alignment[id].seq
 
             for entry in AAs:
@@ -111,7 +112,7 @@ def type_aas_and_dels(in_fasta, in_aa_file, in_del_file, reference_fasta, in_met
 
 def main():
     args = parse_args()
-    type_aas_and_dels(args.in_fasta, args.aas, args.dels, args.reference_fasta, args.in_metadata, args.out_metadata)
+    type_aas_and_dels(args.in_fasta, args.aas, args.dels, args.reference_fasta, args.in_metadata, args.out_metadata, args.index_column)
 
 if __name__ == '__main__':
     main()

modules/align_and_variant_call_cog_uk.nf

@@ -180,19 +180,17 @@ process uk_type_AAs_and_dels {
 workflow align_and_variant_call_cog_uk {
     take:
         uk_fasta
-        uk_metadata
     main:
         uk_minimap2_to_reference(uk_fasta)
         uk_get_variants(uk_minimap2_to_reference.out)
         uk_get_indels(uk_minimap2_to_reference.out)
         uk_alignment(uk_minimap2_to_reference.out)
         uk_mask_alignment(uk_alignment.out)
         uk_get_snps(uk_mask_alignment.out)
-        uk_type_AAs_and_dels(uk_mask_alignment.out, uk_metadata)
+        uk_type_AAs_and_dels(uk_mask_alignment.out, uk_get_variants.out)
     emit:
-        variants = uk_get_variants.out
+        variants = uk_type_AAs_and_dels.out
         fasta = uk_mask_alignment.out
-        metadata = uk_type_AAs_and_dels.out
 }
 
 
@@ -205,8 +203,6 @@ reference_genbank = file(params.reference_genbank)
 
 workflow {
     uk_fasta = Channel.fromPath(params.uk_fasta)
-    uk_metadata = Channel.fromPath(params.uk_metadata)
 
-    align_and_variant_call_cog_uk(uk_fasta,
-                                  uk_metadata)
+    align_and_variant_call_cog_uk(uk_fasta)
 }

modules/deduplicate_cog_uk.nf

@@ -254,7 +254,7 @@ process uk_remove_duplicates_rootbiosample_by_gaps {
 }
 
 
-process unify_headers {
+process uk_unify_headers {
     input:
     path uk_fasta
     path uk_metadata
@@ -275,7 +275,7 @@ process unify_headers {
         reader = csv.DictReader(csv_in, delimiter=",", quotechar='\"', dialect = "unix")
         for row in reader:
             record = alignment[row["fasta_header"]]
-            fasta_out.write(">" + row["sample_name"] + "\\n")
+            fasta_out.write(">" + row["sequence_name"] + "\\n")
             fasta_out.write(str(record.seq) + "\\n")
     """
 }

workflows/process_cog_uk.nf

@@ -19,8 +19,8 @@ workflow process_cog_uk {
       preprocess_cog_uk(uk_fasta, uk_metadata, uk_accessions)
       pangolin_cog_uk(preprocess_cog_uk.out.fasta, preprocess_cog_uk.out.metadata)
       deduplicate_cog_uk(preprocess_cog_uk.out.fasta, pangolin_cog_uk.out.metadata)
-      align_and_variant_call_cog_uk(deduplicate_cog_uk.out.fasta, deduplicate_cog_uk.out.metadata)
-      filter_and_trim_cog_uk(align_and_variant_call_cog_uk.out.fasta, align_and_variant_call_cog_uk.out.metadata)
+      align_and_variant_call_cog_uk(deduplicate_cog_uk.out.fasta)
+      filter_and_trim_cog_uk(align_and_variant_call_cog_uk.out.fasta, deduplicate_cog_uk.out.metadata)
     emit:
       unaligned_fasta = deduplicate_cog_uk.out.metadata
       aligned_fasta = align_and_variant_call_cog_uk.out.fasta