fix: A sample column value of 0|0 is not being parsed correctly

tests/test_call.py

@@ -27,6 +27,19 @@ def build_rec(calls=None, format_extras=None):
  )
 
 
+# Call.__init__() ------------------------------------------------------------
+
+
+def test_call_init_with_gt():
+ call = record.Call("sample", vcfpy.OrderedDict([("GT", "0|1")]))
+ expected_length = 2
+ expected_allele_0 = 0
+ expected_allele_1 = 1
+ assert len(call.gt_alleles) == expected_length
+ assert call.gt_alleles[0] == expected_allele_0
+ assert call.gt_alleles[1] == expected_allele_1
+
+
 # Call.is_phased() ------------------------------------------------------------
 
 

tests/test_header.py

@@ -348,3 +348,25 @@ def test_header_has_header_line_positive_no_samples():
  assert not hdr.has_header_line("INFO", "AD")
  assert not hdr.has_header_line("FILTER", "PASS")
  assert not hdr.has_header_line("contig", "1")
+
+
+def test_header_get_format_field_info():
+ lines = []
+ samples = header.SamplesInfos(["one", "two", "three"])
+ hdr = header.Header(lines, samples)
+ gt_field_info = hdr.get_format_field_info("GT")
+
+ expected = header.RESERVED_FORMAT["GT"]
+
+ assert gt_field_info is expected
+
+
+def test_header_get_info_format_field_info():
+ lines = []
+ samples = header.SamplesInfos(["one", "two", "three"])
+ hdr = header.Header(lines, samples)
+ gt_field_info = hdr.get_info_field_info("AA")
+
+ expected = header.RESERVED_INFO["AA"]
+
+ assert gt_field_info is expected

tests/test_parser_record.py

@@ -10,6 +10,12 @@
 __author__ = "Manuel Holtgrewe <manuel.holtgrewe@bihealth.de>"
 
 
+SMALL_HEADER = """
+##fileformat=VCFv4.3
+#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tNA00001\tNA00002\tNA00003
+""".lstrip()
+
+
 MEDIUM_HEADER = """
 ##fileformat=VCFv4.3
 ##fileDate=20090805
@@ -178,3 +184,13 @@ def test_missing_pass(recwarn):
  RESULT = p.parse_next_record()
  assert str(RESULT) == EXPECTED
  assert list(recwarn) == []
+
+
+def test_parse_record_with_gt_data():
+ LINES = "20\t1\t.\tC\tG\t.\t.\tAA=G\tGT\t0|1\t1/1\t.\n"
+ p = parser.Parser(io.StringIO(SMALL_HEADER + LINES))
+ p.parse_header()
+ record = p.parse_next_record()
+ assert record.calls[0].data["GT"] == "0|1"
+ assert record.calls[1].data["GT"] == "1/1"
+ assert record.calls[2].data["GT"] is None

vcfpy/header.py

@@ -396,18 +396,18 @@ def add_line(self, header_line):
 
  def get_info_field_info(self, key):
  """Return :py:class:`FieldInfo` for the given INFO field"""
- return self._get_field_info("INFO", key)
+ return self._get_field_info("INFO", key, RESERVED_INFO)
 
  def get_format_field_info(self, key):
  """Return :py:class:`FieldInfo` for the given INFO field"""
- return self._get_field_info("FORMAT", key)
+ return self._get_field_info("FORMAT", key, RESERVED_FORMAT)
 
- def _get_field_info(self, type_, key):
+ def _get_field_info(self, type_, key, reserved):
  result = self._indices[type_].get(key)
  if result:
  return result
- if key in RESERVED_INFO:
- res = FieldInfo(RESERVED_INFO[key].type, RESERVED_INFO[key].number)
+ if key in reserved:
+ res = reserved[key]
  else:
  res = FieldInfo("String", HEADER_NUMBER_UNBOUNDED)
  warnings.warn(