CPG-only, remove chrM from small variant process

src/talos/annotation_scripts/annotated_vcf_into_matrixtable.py

@@ -191,7 +191,11 @@ def annotate_all_transcript_consequences(
                     mane[x.transcript]['mane_id'],
                     MISSING_STRING,
                 ),
-                gene_id=ensgs[mt.locus.contig].get(x.gene, x.gene),
+                gene_id=hl.if_else(
+                    ensgs.contains(mt.locus.contig),
+                    ensgs[mt.locus.contig].get(x.gene, x.gene),
+                    x.gene,
+                ),
             ),
             mt.transcript_consequences,
         ),

src/talos/annotation_scripts/create_roi_from_gff3.py

@@ -159,7 +159,7 @@ def cli_main():
     parser.add_argument(
         '--merged_output',
         help='Path to output file, regions merged',
-        required=False,
+        required=True,
     )
     parser.add_argument(
         '--flanking',

src/talos/cpg_internal_scripts/extract_fragmented_vcf_from_mt.py

@@ -54,6 +54,8 @@
     },
 }
 
+CANONICAL_CONTIGS = {f'chr{x}' for x in list(range(1, 23))} | {'chrX', 'chrY'}
+
 
 def filter_mt_to_sgids(
     mt: hl.MatrixTable,
@@ -84,6 +86,12 @@ def filter_mt_to_sgids(
     return mt.filter_rows(hl.agg.any(mt.GT.is_non_ref()))
 
 
+def filter_mt_to_contigs(mt: hl.MatrixTable) -> hl.MatrixTable:
+    """Strict filter to just the non-Mito contigs."""
+    contig_literal = hl.literal(CANONICAL_CONTIGS)
+    return mt.filter_rows(contig_literal.contains(mt.locus.contig))
+
+
 def main(
     mt_path: str,
     sg_id_file: str,
@@ -109,6 +117,9 @@ def main(
     # filter the MT to a specific set of SG IDs
     mt = filter_mt_to_sgids(mt, sg_id_file)
 
+    # filter out chrM in this context
+    mt = filter_mt_to_contigs(mt)
+
     # replace the existing INFO block to just have AC/AN/AF - no other carry-over. Allow for this to be missing.
     if 'AF' not in mt.info:
         mt = hl.variant_qc(mt)