ex-266 (jebene) modified _validate_consistent_samples to take

vcf_readers instead of file_readers

jacquard/merge.py

@@ -658,25 +658,23 @@ def _build_writers_to_readers(vcf_readers, output_path):
 
     return writers_to_readers
 
-def _get_readers_per_patient(file_readers):
+def _get_readers_per_patient(vcf_readers):
     readers_per_patient = defaultdict(list)
-    for file_reader in file_readers:
-        file_reader.open()
-        patient = file_reader.file_name.split(".")[0]
-        for line in file_reader.read_lines():
-            caller_meta_header = "##jacquard.translate.caller"
-            if line.startswith(caller_meta_header):
-                readers_per_patient[patient].append(line.split("=")[1])
-        file_reader.close()
+    for vcf_reader in vcf_readers:
+        patient = vcf_reader.file_name.split(".")[0]
+        for metaheader in vcf_reader.metaheaders:
+            if metaheader.startswith("##jacquard.translate.caller"):
+                readers_per_patient[patient].append(metaheader.split("=")[1])
 
     return OrderedDict(sorted(readers_per_patient.items()))
 
-def _validate_consistent_samples(file_readers):
-    readers_per_patient = _get_readers_per_patient(file_readers)
+def _validate_consistent_samples(vcf_readers):
+    readers_per_patient = _get_readers_per_patient(vcf_readers)
 
     all_callers = set()
     for callers in readers_per_patient.values():
         all_callers.update(callers)
+
     warning = 0
     for patient, callers in readers_per_patient.items():
         missing_callers = set(all_callers).difference(set(callers))
@@ -812,7 +810,6 @@ def execute(args, execution_context):
 
     input_files = sorted(glob.glob(os.path.join(input_path, "*.vcf")))
     file_readers = [vcf.FileReader(i) for i in input_files]
-    _validate_consistent_samples(file_readers)
 
     try:
         file_writer = vcf.FileWriter(output_path)
@@ -824,6 +821,7 @@ def execute(args, execution_context):
 #reduce excess iterations over the coordinates/vcf_readers
         merge_vcf_readers = _create_vcf_readers(file_readers, format_tag_regex)
         _validate_consistent_input(merge_vcf_readers, args.include_all)
+        _validate_consistent_samples(merge_vcf_readers)
         merge_vcf_readers = _sort_readers(merge_vcf_readers,
                                           output_path,
                                           format_tag_regex)

test/merge_test.py

@@ -579,19 +579,19 @@ def test_validate_consistent_arguments(self):
                 self.assertFalse("[{}] not found in {}".format(key, row_help))
 
     def test_validate_consistent_samples_missingCaller(self):
-        input_files = [MockFileReader("A.mutect.vcf",
-                                      ["##jacquard.translate.caller=MuTect"]),
-                       MockFileReader("A.strelka.vcf",
-                                      ["##jacquard.translate.caller=Strelka"]),
-                       MockFileReader("A.varscan.vcf",
-                                      ["##jacquard.translate.caller=VarScan"]),
-                       MockFileReader("B.strelka.vcf",
-                                      ["##jacquard.translate.caller=Strelka"]),
-                       MockFileReader("B.varscan.vcf",
-                                      ["##jacquard.translate.caller=VarScan"]),
-                       MockFileReader("C.varscan.vcf",
-                                      ["##jacquard.translate.caller=VarScan"])]
-        merge._validate_consistent_samples(input_files)
+        vcf_readers = [MockVcfReader(input_filepath="A.mutect.vcf",
+                                     metaheaders=["##jacquard.translate.caller=MuTect"]),
+                       MockVcfReader(input_filepath="A.strelka.vcf",
+                                     metaheaders=["##jacquard.translate.caller=Strelka"]),
+                       MockVcfReader(input_filepath="A.varscan.vcf",
+                                     metaheaders=["##jacquard.translate.caller=VarScan"]),
+                       MockVcfReader(input_filepath="B.strelka.vcf",
+                                     metaheaders=["##jacquard.translate.caller=Strelka"]),
+                       MockVcfReader(input_filepath="B.varscan.vcf",
+                                     metaheaders=["##jacquard.translate.caller=VarScan"]),
+                       MockVcfReader(input_filepath="C.varscan.vcf",
+                                     metaheaders=["##jacquard.translate.caller=VarScan"])]
+        merge._validate_consistent_samples(vcf_readers)
 
         actual_log_warnings = test.utils.mock_logger.messages["WARNING"]
         expected_log_warnings = ["Sample [B] is missing VCF(s): ['MuTect']",
@@ -600,13 +600,13 @@ def test_validate_consistent_samples_missingCaller(self):
         self.assertEquals(expected_log_warnings, actual_log_warnings)
 
     def test_validate_consistent_samples_allMissingCallers(self):
-        input_files = [MockFileReader("A.mutect.vcf",
-                                      ["##jacquard.translate.caller=MuTect"]),
-                       MockFileReader("B.strelka.vcf",
-                                      ["##jacquard.translate.caller=Strelka"]),
-                       MockFileReader("C.mutect.vcf",
-                                      ["##jacquard.translate.caller=MuTect"])]
-        merge._validate_consistent_samples(input_files)
+        vcf_readers = [MockVcfReader(input_filepath="A.mutect.vcf",
+                                     metaheaders=["##jacquard.translate.caller=MuTect"]),
+                       MockVcfReader(input_filepath="B.strelka.vcf",
+                                     metaheaders=["##jacquard.translate.caller=Strelka"]),
+                       MockVcfReader(input_filepath="C.mutect.vcf",
+                                     metaheaders=["##jacquard.translate.caller=MuTect"])]
+        merge._validate_consistent_samples(vcf_readers)
 
         actual_log_warnings = test.utils.mock_logger.messages["WARNING"]
         expected_log_warnings = ["Sample [A] is missing VCF(s): ['Strelka']",