1.6.2 PR [closes #160] (#161)

* Add linearise ref option

* linearise fasta in validator

* triple quotes around awk

* escape newlines

* add space after awk

* Fix the awk

* remove the awkward awk solution

* 1.6.2 version bump

Author: BioWilko

artic/fasta_header.py

@@ -5,12 +5,14 @@ def fasta_header(args):
     with open(args.filename) as fh:
         rec = list(SeqIO.parse(fh, "fasta"))
 
+    fasta_format = "fasta-2line" if args.linearise_fasta else "fasta"
+
     with open(args.filename, "w") as fh:
         for record in rec:
             chrom = record.id
             record.id = f"{args.samplename}/{chrom}/ARTIC/{args.caller}"
 
-            SeqIO.write(record, fh, "fasta")
+            SeqIO.write(record, fh, fasta_format)
 
 
 def main():
@@ -22,6 +24,7 @@ def main():
     parser.add_argument("filename")
     parser.add_argument("samplename")
     parser.add_argument("caller")
+    parser.add_argument("--linearise-fasta", action="store_true")
 
     args = parser.parse_args()
     fasta_header(args)

artic/minion.py

@@ -126,7 +126,7 @@ def run(parser, args):
     ## create a holder to keep the pipeline commands in
     cmds = []
 
-    # 2) check the reference fasta has and index and create one if not
+    # 2) check the reference fasta has an index and create one if not
     if not os.path.exists("%s.fai" % (ref)):
         cmds.append("samtools faidx %s" % (ref))
 
@@ -234,8 +234,9 @@ def run(parser, args):
 
     # 11) apply the header to the consensus sequence and run alignment against the reference sequence
     caller = "clair3"
+    linearise_fasta = "--linearise-fasta" if args.linearise_fasta else ""
     cmds.append(
-        f"artic_fasta_header {args.sample}.consensus.fasta {args.sample} {caller}"
+        f"artic_fasta_header {linearise_fasta} {args.sample}.consensus.fasta {args.sample} {caller}"
     )
 
     if args.align_consensus:

artic/pipeline.py

@@ -176,6 +176,11 @@ def init_pipeline_parser():
         action="store_true",
         help="Run a mafft alignment of consensus to reference after generation",
     )
+    parser_minion.add_argument(
+        "--linearise-fasta",
+        action="store_true",
+        help="Output linearised (unwrapped) FASTA consensus files",
+    )
     parser_minion.add_argument("--dry-run", action="store_true")
     parser_minion.set_defaults(func=run_subtool)
 

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = artic
-version = 1.6.1
+version = 1.6.2
 author = Nick Loman
 author_email = [email protected]
 maintainer = Sam Wilkinson

tests/minion_validator.py

@@ -179,6 +179,7 @@ def genCommand(sampleID, workflow):
         "minion",
         "--threads",
         "2",
+        "--linearise-fasta",
         "--read-file",
         dataDir + sampleID + "/" + sampleID + ".fastq",
         "--scheme-directory",