output original AO

src/scanitd/base/basic.py

@@ -54,6 +54,7 @@ class Event:
     __slots__ = (
         "af",
         "alt_allele",
+        "oao",
         "ao",
         "break_point_region",
         "chrom",
@@ -74,6 +75,7 @@ def __init__(
         event_size: int,
         event_sequence: str,
         event_type: str,
+        oao: int,
         ao: int,
         dp: int,
         end: int,
@@ -86,6 +88,7 @@ def __init__(
         self.event_size = event_size
         self.event_sequence = event_sequence
         self.event_type = event_type
+        self.oao = oao
         self.ao = ao
         self.dp = dp
         self.af = round(float(ao / dp), 4)
@@ -114,6 +117,7 @@ def new(
         cls,
         event_type: str,
         event_id: tuple[str, int, int, str, MicroRegion],
+        oao: int,
         ao: int,
         dp: int,
         ref_allele: str | None = None,
@@ -132,6 +136,7 @@ def new(
             event_size,
             event_sequence,
             event_type,
+            oao,
             ao,
             dp,
             end,

src/scanitd/inference/main.py

@@ -383,14 +383,14 @@ def scan_itd(
         _chrom, _ref_start, _event_size, _event_seq, break_point_region = tdup_id
         depth = obtain_depth_given_genomic_position(bam_object, _chrom, _ref_start)
         ref_allele, alt_allele = tdup_allele_dict[tdup_id]
-        event_list.append(Event.new("TDUP", tdup_id, new_ao, depth, ref_allele, alt_allele))
+        event_list.append(Event.new("TDUP", tdup_id, original_ao, new_ao, depth, ref_allele, alt_allele))
 
     for ins_id in ins_ao:
         _chrom, _ref_start, _event_size, _event_seq, break_point_region = ins_id
         ao = ins_ao[tdup_id]
         depth = obtain_depth_given_genomic_position(bam_object, _chrom, _ref_start)
         ref_allele, alt_allele = ins_allele_dict[ins_id]
-        event_list.append(Event.new("INS", ins_id, ao, depth, ref_allele, alt_allele))
+        event_list.append(Event.new("INS", ins_id, ao, ao, depth, ref_allele, alt_allele))
 
     # Sort by chrom, then by reference position
     sorted_event_list = sorted(event_list, key=lambda event: (event.chrom, event.ref_start))

src/scanitd/writer/vcf_writer.py

@@ -45,6 +45,7 @@ class VCFWriter(Writer):
 
     reserved_info: ClassVar[dict[str, str]] = {
         "DP": "Integer",
+        "OAO": "Integer",
         "AO": "Integer",
         "AF": "Float",
         "SVMETHOD": "String",
@@ -64,7 +65,8 @@ class VCFWriter(Writer):
 
     description: ClassVar[dict[str, str]] = {
         "DP": "Total read depth at the locus",
-        "AO": "Alternate allele observations, with partial observations recorded fractionally",
+        "OAO": "Original alternate allele observations",
+        "AO": "Alternate allele observations",
         "AF": "Estimated allele frequency in the range (0,1], representing the ratio of reads showing the alternative allele to all reads",
         "SVTYPE": "The type of event, TDUP, INS.",
         "SVLEN": "Difference in length between REF and ALT alleles",
@@ -226,6 +228,7 @@ def get_vcf_features_from_event(
     sv_type = event.event_type
     chrom = event.chrom
     ref_start = event.ref_start
+    oao = event.oao
     ao = event.ao
     dp = event.dp
     af = event.af
@@ -248,6 +251,7 @@ def get_vcf_features_from_event(
         "SVTYPE": sv_type,
         "CHR2": chrom,
         "END": f"{end + 1}",
+        "OAO": f"{oao}",
         "AO": f"{ao}",
         "DP": f"{dp}",
         "AF": f"{af:.3g}",
@@ -262,7 +266,7 @@ def get_vcf_features_from_event(
 def vcf_feature_transformer(feature_dict: dict[str, str], event_id: int) -> list[str]:
     """VCF feature transformer."""
     info_field = (
-        f'SVTYPE={feature_dict["SVTYPE"]};AO={feature_dict["AO"]};'
+        f'SVTYPE={feature_dict["SVTYPE"]};OAO={feature_dict["OAO"]};AO={feature_dict["AO"]};'
         f'CHR2={feature_dict["CHR2"]};END={feature_dict["END"]};'
         f'DP={feature_dict["DP"]};AF={feature_dict["AF"]};SVLEN={feature_dict["SVLEN"]};'
         f'INSSEQ={feature_dict["INSSEQ"]};HOMSEQ={feature_dict["HOMSEQ"]};'