EX-189 (cgates): Actually enabled sorting in merge

umich-brcf-bioinf · Mar 2, 2015 · 06a686c · 06a686c
1 parent 5d08d1a
commit 06a686c
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 31 deletions.
diff --git a/jacquard/merge.py b/jacquard/merge.py
@@ -110,32 +110,19 @@ def _create_reader_lists(input_files):
 def _build_coordinates(vcf_readers):
     coordinate_set = set()
     mult_alts = defaultdict(set)
-    error = 0
 
     for vcf_reader in vcf_readers:
-        previous_record = None
         try:
             vcf_reader.open()
 
             for vcf_record in vcf_reader.vcf_records():
-                if previous_record and vcf_record < previous_record:
-                    logger.error("VCF file:chrom:pos [{}:{}:{}] is out of order"
-                                 .format(vcf_reader.file_name,
-                                         vcf_record.chrom,
-                                         vcf_record.pos))
-                    error = 1
-                previous_record = vcf_record
                 coordinate_set.add(vcf_record.get_empty_record())
                 ref_alt = vcf_record.ref, vcf_record.alt
                 locus = vcf_record.chrom, vcf_record.pos
                 mult_alts[locus].add(ref_alt)
         finally:
             vcf_reader.close()
 
-    if error:
-        raise utils.JQException("One or more VCF files were not sorted. "
-                                "Review inputs and try again.")
-
     for vcf_record in coordinate_set:
         ref_alts_for_this_locus = mult_alts[vcf_record.chrom,
                                             vcf_record.pos]
@@ -185,6 +172,10 @@ def _get_unsorted_readers(vcf_readers):
         reader.open()
         for vcf_record in reader.vcf_records():
             if previous_record and vcf_record < previous_record:
+                logger.debug("VCF file:chrom:pos [{}:{}:{}] is out of order"
+                             .format(reader.file_name,
+                                     vcf_record.chrom,
+                                     vcf_record.pos))
                 unsorted_readers.append(reader)
                 break
             else:
@@ -379,6 +370,7 @@ def execute(args, execution_context):
 
         buffered_readers, vcf_readers = _create_reader_lists(input_files)
 
+        vcf_readers = _sort_readers(vcf_readers, output_path)
         all_sample_names, merge_metaheaders = _build_sample_list(vcf_readers)
         coordinates = _build_coordinates(vcf_readers)
         format_tags_to_keep = _build_format_tags(format_tag_regex, vcf_readers)

diff --git a/test/merge_test.py b/test/merge_test.py
@@ -93,24 +93,6 @@ def test_build_coordinates(self):
         expected = [fileArec1, fileArec2, fileBrec2]
         self.assertEquals(expected, actual_coordinates)
 
-    def test_build_coordinates_unsorted(self):
-        fileArec1 = vcf.VcfRecord("chr1", "1", "A", "C")
-        fileArec2 = vcf.VcfRecord("chr2", "5", "A", "G", "id=1")
-        fileBrec1 = vcf.VcfRecord("chr2", "16", "A", "G", "id=2")
-        fileBrec2 = vcf.VcfRecord("chr2", "12", "G", "C")
-
-        mock_readers = [MockVcfReader(input_filepath="fileA.vcf",
-                                      records=[fileArec1, fileArec2]),
-                        MockVcfReader(input_filepath="fileB.vcf",
-                                      records=[fileBrec1, fileBrec2])]
-
-        self.assertRaisesRegexp(utils.JQException,
-                                "One or more VCF files were not sorted.*",
-                                merge._build_coordinates, mock_readers)
-        actual_log_errors = test.mock_logger.messages["ERROR"]
-        self.assertRegexpMatches(actual_log_errors[0],
-                                 r"VCF file:chrom:pos \[fileB.vcf:chr2:12\] is out of order")
-
     def test_build_coordinates_multAltsEmpty(self):
         fileArec1 = vcf.VcfRecord("chr1", "1", "A", "C")
         fileArec2 = vcf.VcfRecord("chr2", "12", "A", "G", "id=1")
@@ -758,6 +740,11 @@ def test_sort_readers_vcfsResortedAsNecessary(self):
         actual_log_infos = test.mock_logger.messages["INFO"]
         self.assertEquals(1, len(actual_log_infos))
         self.assertRegexpMatches(actual_log_infos[0], r"Sorting vcf \[unsorted.vcf\]")
+        actual_log_debugs = test.mock_logger.messages["DEBUG"]
+        self.assertEquals(1, len(actual_log_debugs))
+        self.assertRegexpMatches(actual_log_debugs[0],
+                                 r"VCF file:chrom:pos \[unsorted.vcf:chr1:42\] is out of order")
+
 
 class MergeFunctionalTestCase(test_case.JacquardBaseTestCase):
     def test_merge(self):