ex-289 (cgates): adjusted varscan validation to allow empty HC files.…

… Some pep8 cleanup.
umich-brcf-bioinf · Sep 2, 2015 · c0d6f2e · c0d6f2e
1 parent 27a3ab2
commit c0d6f2e
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 7 deletions.
diff --git a/jacquard/variant_caller_transforms/varscan.py b/jacquard/variant_caller_transforms/varscan.py
@@ -201,13 +201,15 @@ def _get_hc_file_pattern(args):
     def _validate_filter_file(file_reader):
         column_header = 0
         file_reader.open()
+        file_is_empty = True
         for line in file_reader.read_lines():
+            file_is_empty = False
             if line.startswith("chrom\tposition"):
                 column_header = line
                 break
         file_reader.close()
 
-        if column_header:
+        if column_header or file_is_empty:
             return file_reader
 
     @staticmethod

diff --git a/test/expand_test.py b/test/expand_test.py
@@ -255,7 +255,7 @@ def test_validate_args_colSpecValid(self):
             input_file = os.path.join(input_dir.path, "input.vcf")
             input_dir.write("col_spec.txt", b"chrom\npos\ninfo")
             col_spec_file = os.path.join(input_dir.path, "col_spec.txt")
- 
+
             args = Namespace(input=input_file,
                              output="expanded.txt",
                              selected_columns_file=col_spec_file)
@@ -286,7 +286,7 @@ def test_validate_args_colSpecIsEmpty(self):
                                     "The selected_columns_file .* has no rows. Review inputs/usage and try again",
                                     expand.validate_args,
                                     args)
-    
+
     def test_validate_args_checkInputIfVCF(self):
         with TempDirectory() as input_dir:
             input_dir.write("input.vcf", b"123")

diff --git a/test/merge_test.py b/test/merge_test.py
@@ -160,13 +160,13 @@ def test_init_includeAllFlag(self):
 #                                 "Unable to process command-line arguments. Neither --include_cells nor --include_rows can be specified if --include_all is specified.",
 #                                  merge._Filter,
 #                                  args)
-# 
+#
 #         args = Namespace(include_all=True, include_cells=False, include_rows=True)
 #         self.assertRaisesRegexp(utils.UsageError,
 #                                 "Unable to process command-line arguments. Neither --include_cells nor --include_rows can be specified if --include_all is specified.",
 #                                  merge._Filter,
 #                                  args)
-# 
+#
 #         args = Namespace(include_all=True, include_cells=True, include_rows=False)
 #         self.assertRaisesRegexp(utils.UsageError,
 #                                 "Unable to process command-line arguments. Neither --include_cells nor --include_rows can be specified if --include_all is specified.",

diff --git a/test/variant_caller_transforms/varscan_test.py b/test/variant_caller_transforms/varscan_test.py
@@ -207,15 +207,15 @@ def test_validate_vcf_hc_pairs_raisesIfMissingHcFiles(self):
         vcf_hc_pairs = [(MockFileReader("A.vcf"), MockFileReader("A.hc")),
                         (MockFileReader("B.vcf"), None)]
         self.assertRaisesRegexp(utils.UsageError,
-                                "The VarScan VCF file \[B.vcf\] has no matching high-confidence file.", 
+                                "The VarScan VCF file \[B.vcf\] has no matching high-confidence file.",
                                 self.caller._validate_vcf_hc_pairs,
                                 vcf_hc_pairs)
 
     def test_validate_vcf_hc_pairs_raisesIfNoMissingVcfFiles(self):
         vcf_hc_pairs = [(MockFileReader("A.vcf"), MockFileReader("A.hc")),
                         (None, MockFileReader("B.hc"))]
         self.assertRaisesRegexp(utils.UsageError,
-                                "The VarScan high-confidence file \[B.hc\] has no matching VCF file.", 
+                                "The VarScan high-confidence file \[B.hc\] has no matching VCF file.",
                                 self.caller._validate_vcf_hc_pairs,
                                 vcf_hc_pairs)
 
@@ -232,6 +232,12 @@ def test_get_hc_file_pattern_invalidRegex(self):
                                 varscan.Varscan._get_hc_file_pattern,
                                 args)
 
+    def test_validate_filter_file_emptyFile(self):
+        file_reader = MockFileReader("p1.hc.fpfilter.pass", [])
+        caller = varscan.Varscan()
+        valid_reader = caller._validate_filter_file(file_reader)
+        self.assertEquals("p1.hc.fpfilter.pass", valid_reader.file_name)
+
     def test_validate_filter_file_validFile(self):
         file_reader = MockFileReader("p1.hc.fpfilter.pass", ["chrom\tposition"])
         caller = varscan.Varscan()