JQ-325 (cgates): Fix sporadically failing (py3) vcf test

umich-brcf-bioinf · Jul 11, 2019 · 322633b · 322633b
1 parent b66d216
commit 322633b
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 9 deletions.
diff --git a/jacquard/utils/vcf.py b/jacquard/utils/vcf.py
@@ -353,11 +353,18 @@ def _sample_field(self, tag_names, sample):
         sample_tag_values = self.sample_tag_values[sample]
         missing_tag_names = set(sample_tag_values) - set(tag_names)
         if missing_tag_names:
-            msg = ('sample format tags are not consistent: '
-                  '{}:{}:{}:{}').format(self.chrom,
-                                        self.pos,
-                                        self.ref,
-                                        self.alt)
+            msg = ('{}:{}:{}:{}|sample format tags are not consistent: '
+                   'requested tags [{}] but sample {} has has tags [{}] '
+                   'leaving behind [{}]')\
+                 .format(self.chrom,
+                         self.pos,
+                         self.ref,
+                         self.alt,
+                         ', '.join(tag_names),
+                         sample,
+                         ', '.join(['{}={}'.format(k,v) for k,v in sample_tag_values.items()]),
+                         ', '.join(missing_tag_names)
+                         )
             raise ValueError(msg)
         tag_values = [sample_tag_values.get(t, '.') for t in tag_names]
         if tag_values:

diff --git a/test/utils/vcf_test.py b/test/utils/vcf_test.py
@@ -525,13 +525,26 @@ def test_asTextWhenEmptyFormatField(self):
         self.assertEquals(expected, record.text())
 
     def test_asTextExpandsEmptyTrailingFormatField(self):
-        sampleA = OrderedDict({'a':'1', 'b':'2'})
-        sampleB = OrderedDict({'a':'10'})
-        sample_tag_values = OrderedDict({"SampleA":sampleA, "SampleB":sampleB})
+        sampleA = OrderedDict([('a','1'), ('b','2')])
+        sampleB = OrderedDict([('a','10')])
+        sample_tag_values = OrderedDict([("SampleA", sampleA), ("SampleB", sampleB)])
         record = VcfRecord("CHROM", "POS", "REF", "ALT", "ID", "QUAL", "FILTER", "INFO", sample_tag_values)
         expected = self.entab("CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|a:b|1:2|10:.\n")
         self.assertEquals(expected, record.text())
 
+    def test_sample_field_whenInconsistentTags(self):
+        # FYI this should never happen in the wild, but I wanted to test the exception formatting.
+        sampleA = OrderedDict([('a','1'), ('b','2')])
+        sampleB = OrderedDict([('a','10')])
+        sample_tag_values = OrderedDict([("SampleA", sampleA), ("SampleB", sampleB)])
+        record = VcfRecord("CHROM", "POS", "REF", "ALT", "ID", "QUAL", "FILTER", "INFO", sample_tag_values)
+
+        self.assertRaisesRegexp(ValueError,
+            r'CHROM:POS:REF:ALT|sample format tags are not consistent: requested tags \[a\] but sample has has tags \[a=1, b=2\] leaving behind \[b\]',
+            record._sample_field,
+            ['a'],
+            'SampleA')
+
 
     def test_equals(self):
         sample_names = ["sampleA"]
@@ -550,7 +563,7 @@ def test_equals(self):
     def testHash(self):
         sample_names = ["sampleA"]
         base = VcfRecord.parse_record(self.entab("A|B|ID|C|D|QUAL|FILTER|INFO|F|S\n"), sample_names)
-        base_equivalent = VcfRecord.parse_record(self.entab("A|B|ID|C|D|QUAL|FILTER||F|S\n"), sample_names)
+        base_equivalent = VcfRecord.parse_record(self.entab("A|B|ID|C|D|QUAL|FILTER||foo|S\n"), sample_names)
         self.assertEquals(base.__hash__(), base_equivalent.__hash__())
         record_set = set()
         record_set.add(base)