test add sage metrics for Msi calculation

Author: qclayssen

bolt/common/constants.py

@@ -77,6 +77,11 @@ class VcfInfo(enum.Enum):
     SAGE_NOVEL = 'SAGE_NOVEL'
     SAGE_RESCUE = 'SAGE_RESCUE'
 
+    RC_REPC = 'RC_REPC'
+    RC_REPS = 'RC_REPS'
+    REP_C = 'REP_C'
+    REP_S = 'REP_S'
+
     PCGR_TIER = 'PCGR_TIER'
     PCGR_CSQ = 'PCGR_CSQ'
     PCGR_MUTATION_HOTSPOT = 'PCGR_MUTATION_HOTSPOT'
@@ -226,6 +231,27 @@ def namespace(self):
         'Description': 'Variant rescued by a matching SAGE call',
     },
 
+    VcfInfo.RC_REPC: {
+        'Number': '1',
+        'Type': 'Integer',
+        'Description': 'Repeat count from read context',
+    },
+    VcfInfo.RC_REPS: {
+        'Number': '1',
+        'Type': 'String',
+        'Description': 'Repeat sequence from read context',
+    },
+    VcfInfo.REP_C: {
+        'Number': '1',
+        'Type': 'Integer',
+        'Description': 'Repeat sequence count',
+    },
+    VcfInfo.REP_S: {
+        'Number': '1',
+        'Type': 'String',
+        'Description': 'Repeat sequence',
+    },
+
     VcfInfo.PCGR_TIER: {
         'Number': '1',
         'Type': 'String',

bolt/workflows/smlv_somatic/rescue.py

@@ -112,6 +112,11 @@ def annotate_existing_sage_calls(input_fp, tumor_name, sage_vcf_fp, output_dir):
     util.add_vcf_header_entry(input_fh, constants.VcfInfo.SAGE_HOTSPOT)
     util.add_vcf_header_entry(input_fh, constants.VcfInfo.SAGE_RESCUE)
 
+    util.add_vcf_header_entry(input_fh, constants.VcfInfo.RC_REPC)
+    util.add_vcf_header_entry(input_fh, constants.VcfInfo.RC_REPS)
+    util.add_vcf_header_entry(input_fh, constants.VcfInfo.REP_C)
+    util.add_vcf_header_entry(input_fh, constants.VcfInfo.REP_S)
+
     # TODO(SW): check that defined header descriptions match those in the SAGE fp; collect as list
     # here and iterate to check and then add to input_fp header also in another loop
 
@@ -155,6 +160,16 @@ def annotate_existing_sage_calls(input_fp, tumor_name, sage_vcf_fp, output_dir):
         record.set_format(constants.VcfFormat.SAGE_DP.value, sage_record.format('DP'))
         record.set_format(constants.VcfFormat.SAGE_SB.value, sage_record.format('SB'))
 
+        # Transfer SAGE INFO fields for repeat context
+        if sage_record.INFO.get('RC_REPC') is not None:
+            record.INFO[constants.VcfInfo.RC_REPC.value] = sage_record.INFO['RC_REPC']
+        if sage_record.INFO.get('RC_REPS') is not None:
+            record.INFO[constants.VcfInfo.RC_REPS.value] = sage_record.INFO['RC_REPS']
+        if sage_record.INFO.get('REP_C') is not None:
+            record.INFO[constants.VcfInfo.REP_C.value] = sage_record.INFO['REP_C']
+        if sage_record.INFO.get('REP_S') is not None:
+            record.INFO[constants.VcfInfo.REP_S.value] = sage_record.INFO['REP_S']
+
         output_fh.write_record(record)
 
     # Explicitly close to flush buffer then index output file
@@ -196,6 +211,10 @@ def prepare_sage_novel(input_fp, tumor_name, output_dir):
     annotations_retain_info = (
         util.get_qualified_vcf_annotation(constants.VcfInfo.SAGE_HOTSPOT),
         util.get_qualified_vcf_annotation(constants.VcfInfo.SAGE_NOVEL),
+        util.get_qualified_vcf_annotation(constants.VcfInfo.RC_REPC),
+        util.get_qualified_vcf_annotation(constants.VcfInfo.RC_REPS),
+        util.get_qualified_vcf_annotation(constants.VcfInfo.REP_C),
+        util.get_qualified_vcf_annotation(constants.VcfInfo.REP_S),
     )
     annotations_retain_format = (
         util.get_qualified_vcf_annotation(constants.VcfFormat.SAGE_SB),
@@ -215,6 +234,10 @@ def prepare_sage_novel(input_fp, tumor_name, output_dir):
     with header_entries_fp.open('w') as fh:
         print(util.get_vcf_header_line(constants.VcfInfo.SAGE_HOTSPOT), file=fh)
         print(util.get_vcf_header_line(constants.VcfInfo.SAGE_NOVEL), file=fh)
+        print(util.get_vcf_header_line(constants.VcfInfo.RC_REPC), file=fh)
+        print(util.get_vcf_header_line(constants.VcfInfo.RC_REPS), file=fh)
+        print(util.get_vcf_header_line(constants.VcfInfo.REP_C), file=fh)
+        print(util.get_vcf_header_line(constants.VcfInfo.REP_S), file=fh)
 
     # NOTE(SW): the fill-tags BCFtools plugin would have been useful here instead of iterating with
     # awk but unforunately it doesn't seem to support anything other than integer and float INFO