1.7.3 PR

artic/fasta_header.py

@@ -1,16 +1,21 @@
 from Bio import SeqIO
+from importlib.metadata import version
 
 
 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"
+    artic_version = version("artic")
 
     with open(args.filename, "w") as fh:
         for record in rec:
             chrom = record.id
-            record.id = f"{args.samplename}/{chrom}/ARTIC/{args.caller}"
+            record.id = f"{args.samplename}"
+            record.description = (
+                f"{chrom}_artic-network/fieldbioinformatics_{artic_version}"
+            )
 
             SeqIO.write(record, fh, fasta_format)
 
@@ -23,7 +28,6 @@ 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()

artic/minion.py

@@ -121,7 +121,7 @@ def run(parser, args):
         raise SystemExit(1)
 
     ## collect the primer pools
-    pools = set([row["PoolName"] for row in read_bed_file(bed)])
+    pools = set([row["PoolName"] for row in read_bed_file(bed)] + ["unmatched"])
 
     ## create a holder to keep the pipeline commands in
     cmds = []
@@ -171,20 +171,19 @@ def run(parser, args):
             os.remove("%s.%s.hdf" % (args.sample, p))
 
         # Split the BAM by read group
-        for p in pools:
-            cmds.append(
-                f"samtools view -b -r {p} {args.sample}.trimmed.rg.sorted.bam -o {args.sample}.{p}.trimmed.rg.sorted.bam"
-            )
+        cmds.append(
+            f"samtools view -b -r {p} {args.sample}.trimmed.rg.sorted.bam -o {args.sample}.{p}.trimmed.rg.sorted.bam"
+        )
 
-            cmds.append(f"samtools index {args.sample}.{p}.trimmed.rg.sorted.bam")
+        cmds.append(f"samtools index {args.sample}.{p}.trimmed.rg.sorted.bam")
 
-            cmds.append(
-                f"run_clair3.sh --enable_long_indel --chunk_size=10000 --haploid_precise --no_phasing_for_fa --bam_fn='{args.sample}.{p}.trimmed.rg.sorted.bam' --ref_fn='{ref}' --output='{args.sample}_rg_{p}' --threads='{args.threads}' --platform='ont' --model_path='{full_model_path}' --include_all_ctgs"
-            )
+        cmds.append(
+            f"run_clair3.sh --enable_long_indel --chunk_size=10000 --haploid_precise --no_phasing_for_fa --bam_fn='{args.sample}.{p}.trimmed.rg.sorted.bam' --ref_fn='{ref}' --output='{args.sample}_rg_{p}' --threads='{args.threads}' --platform='ont' --model_path='{full_model_path}' --include_all_ctgs"
+        )
 
-            cmds.append(
-                f"bgzip -dc {args.sample}_rg_{p}/merge_output.vcf.gz > {args.sample}.{p}.vcf"
-            )
+        cmds.append(
+            f"bgzip -dc {args.sample}_rg_{p}/merge_output.vcf.gz > {args.sample}.{p}.vcf"
+        )
 
     # 7) merge the called variants for each read group
     merge_vcf_cmd = "artic_vcf_merge %s %s 2> %s.primersitereport.txt" % (
@@ -237,7 +236,7 @@ def run(parser, args):
     caller = "clair3"
     linearise_fasta = "--linearise-fasta" if args.linearise_fasta else ""
     cmds.append(
-        f"artic_fasta_header {linearise_fasta} {args.sample}.consensus.fasta {args.sample} {caller}"
+        f"artic_fasta_header {linearise_fasta} {args.sample}.consensus.fasta {args.sample}"
     )
 
     if args.align_consensus:

tests/minion_validator.py

@@ -123,13 +123,11 @@
 
 # extraFlags is a way to add extra sample-specific commands to the validation test cmd
 extraFlags = {
-    "medaka": {
-        "SP1": ["--no-frameshifts"],
-    },
     "clair3": {
         "SP1": ["--no-frameshifts"],
         "CVR1": ["--no-frameshifts"],
     },
+    "clair3_allow_mismatches": {"CVR1": ["--allow-mismatched-primers"]},
 }
 
 
@@ -186,7 +184,7 @@ def genCommand(sampleID, workflow):
         dataDir + "primer-schemes",
     ]
 
-    if workflow == "clair3":
+    if workflow == "clair3" or workflow == "clair3_allow_mismatches":
         cmd.append("--model")
         cmd.append("r941_prom_hac_g360+g422")
 
@@ -225,6 +223,8 @@ def runner(workflow, sampleID):
 
     if workflow == "clair3":
         data = clair3TestVariants
+    elif workflow == "clair3_allow_mismatches":
+        data = clair3TestVariants
     elif workflow == "medaka":
         data = medakaTestVariants
     else:
@@ -355,6 +355,9 @@ class TestMinion(unittest.TestCase):
     def setUp(self):
         dataChecker()
 
+    def test_Clair3_CVR1_allow_primer_mismatches(self):
+        runner("clair3_allow_mismatches", "CVR1")
+
     def test_Clair3_CVR1(self):
         runner("clair3", "CVR1")