add 20A-I

sanger-bentley-group · Jul 22, 2024 · 96aa780 · 96aa780
1 parent 346051c
commit 96aa780
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 1 deletion.
diff --git a/database/meta.tsv b/database/meta.tsv
@@ -70,6 +70,7 @@
 19FIII	19F	19FIII	1	ref
 19FIV	19F	19FIV	1	ref
 20A	20A	20A	1	ref
+20AI	20A	20AI	1	ref
 20B	20A	20B	1	ref
 21	21	21	1	ref
 22A	22A	22A	1	ref

diff --git a/database/reference.fasta b/database/reference.fasta
diff --git a/database/streptococcus-pneumoniae-ctvdb/reference.fasta b/database/streptococcus-pneumoniae-ctvdb/reference.fasta
diff --git a/seroba/serotyping.py b/seroba/serotyping.py
@@ -604,6 +604,14 @@ def run(self):
                 if '24B' in self.sero or '24F' in self.sero or '24C' in self.sero:
                     fobj.write(header)
                     fobj.write(f"{self.prefix},24B/24C/24F,24B/24C/24F,{flag}\n")
+                elif '20' in self.sero:
+                    # early stop codon in whaF leads to no assembly of whaF, check this via file size comparison
+                    gene_size = os.path.getsize(f"{self.prefix}/assembled_genes.fa")
+                    fobj.write(header)
+                    if gene_size == 0:
+                        fobj.write(f"{self.prefix},20A,20A(20A-I),{flag}\n")
+                    else:
+                        fobj.write(f"{self.prefix},{self.sero},{self.sero},{flag}\n")
                 elif 'possible' in self.sero:
                     # catch uncertain serogroup 6 calls
                     fobj.write(header)