ex-264 (cgates): Adjusted to work in py3; minor PEP8 cleanup; removed…

… some deprecated methods

jacquard/expand.py

@@ -192,7 +192,9 @@ def execute(args, dummy_execution_context):
     #for the moment, there is no good place to put the execution context
     input_file = os.path.abspath(args.input)
     output_file = os.path.abspath(args.output)
-    col_spec = args.selected_columns_file if args.selected_columns_file else None
+    col_spec = None
+    if args.selected_columns_file:
+        col_spec = args.selected_columns_file
 
     logger.debug("Expanding [{}] to [{}]",
                  input_file,

jacquard/translate.py

@@ -250,8 +250,8 @@ def add_subparser(subparser):
     parser.add_argument("-v", "--verbose", action='store_true')
 
 #TODO (cgates): This module is both a command and also manipulates VcfRecords
-# like a caller. This is the only body of code that does both these things.
-# Does this bother anyone else?
+# like a caller transform. This is the only body of code that does both these
+# things. Does this bother anyone else?
 def execute(args, execution_context):
     validate_args(args)
 

jacquard/utils/summarize_rollup_transform.py

@@ -228,6 +228,7 @@ def add_tag_values(vcf_record):
         _add_sample_count_values(vcf_record, JQ_PASSED, JQ_SAMPLES_PASSED)
 
 class _HCGenotypeTag(object):
+    #pylint: disable=too-few-public-methods
     _TAG_ID = "{}HC_GT".format(JQ_SUMMARY_TAG)
     _PATTERN = re.compile("^JQ_.*_GT$")
 

jacquard/utils/summarize_zscore_transform.py

@@ -163,7 +163,7 @@ def _get_dependent_value(tag_values, dependent_tag_id):
             return None
 
     def _init_population_stats(self, vcf_reader, dependent_tag_id):
-        '''Derive mean and stdev. 
+        '''Derive mean and stdev.
 
         Adapted from online variance algorithm from Knuth, The Art of Computer 
         Programming, volume 2

jacquard/utils/utils.py

@@ -40,6 +40,7 @@ def sort_metaheaders(metaheaders):
                                                          .split("=")[0]])
 
 class _JacquardHelpFormatter(argparse.RawTextHelpFormatter):
+    #pylint: disable=too-few-public-methods
     def _format_usage(self, default_values):
         prog = '%(prog)s' % dict(prog=self._prog)
         usage = 'usage: {} <input> <output> {}'.format(prog, default_values)
@@ -48,6 +49,7 @@ def _format_usage(self, default_values):
     def add_usage(self, default_values, actions=None, groups=None, prefix=None):
         self._add_item(self._format_usage, default_values)
 
+
 class JQException(Exception):
     """Base class for all run-time exceptions in this module."""
     def __init__(self, msg, *args):

jacquard/variant_caller_transforms/common_tags.py

@@ -1,7 +1,6 @@
 """Common tags used by several callers."""
 from __future__ import print_function, absolute_import, division
 import abc
-from abc import abstractmethod
 import jacquard.utils.utils as utils
 
 CALLER_REPORTED_TAG = "CALLER_REPORTED"
@@ -24,24 +23,21 @@ def __init__(self, abbreviation, vcf_type, vcf_number):
     GENOTYPE_TAG = _TagType("GT", "String", "1")
     ALLELE_FREQ_TAG = _TagType("AF", "Float", "A")
     SOMATIC_TAG = _TagType("HC_SOM", "Integer", "1")
-    
+
     def __init__(self, variant_caller_abbrev, tag_type, description):
         if '"' in description:
             raise utils.JQException(("Metaheader descriptions cannot contain "
                                     "double quotes: [{}]"),
                                     description)
-        self.tag_id = "JQ_{}_{}".format(variant_caller_abbrev,tag_type.abbreviation)
-
+        self.tag_id = "JQ_{}_{}".format(variant_caller_abbrev,
+                                        tag_type.abbreviation)
         self.metaheader = JacquardTag.FORMAT.format(self.tag_id,
                                                     tag_type.vcf_number,
                                                     tag_type.vcf_type,
                                                     description)
-
-
 
-    @abstractmethod
-    def add_tag_values(self):
-        pass
+    def add_tag_values(self, vcf_record):
+        raise NotImplementedError()
 
 
 class ReportedTag(object):

jacquard/variant_caller_transforms/mutect.py

@@ -76,12 +76,15 @@ def add_tag_values(self, vcf_record):
             vcf_record.add_sample_tag_value(self.tag_id, sample_values)
 
 class _SomaticTag(common_tags.JacquardTag):
+    _DESCRIPTION = (\
+'''Jacquard somatic status for MuTect: 0=non-somatic,1=somatic (based on SS 
+FORMAT tag)''').replace("\n","")
     #pylint: disable=too-few-public-methods
     def __init__(self):
         super(self.__class__,
               self).__init__(MUTECT_ABBREVIATION,
                              common_tags.JacquardTag.SOMATIC_TAG,
-                             'Jacquard somatic status for MuTect: 0=non-somatic,1=somatic (based on SS FORMAT tag)')
+                             self._DESCRIPTION)
 
     def add_tag_values(self, vcf_record):
         sample_values = {}
@@ -117,14 +120,6 @@ def __init__(self):
         self.name = "MuTect"
         self.abbr = "MT"
 
-    ##TODO (cgates): deprecate; remove
-    @staticmethod
-    def validate_input_file(meta_headers, dummy_column_header):
-        for line in meta_headers:
-            if line.startswith(Mutect._MUTECT_METAHEADER_PREFIX):
-                return True
-        return False
-
     @staticmethod
     def _is_mutect_vcf(file_reader):
         if file_reader.file_name.lower().endswith(".vcf"):

jacquard/variant_caller_transforms/strelka.py

@@ -26,6 +26,8 @@
 VERSION = "v2.0.15"
 
 class _GenotypeTag(common_tags.JacquardTag):
+    #pylint: disable=too-few-public-methods
+
     _INDEL_VALUES = ["ref", "hom", "het"]
 
     def __init__(self):
@@ -231,11 +233,6 @@ def __init__(self):
         self.abbr = "SK"
         self.meta_header = "##jacquard.normalize_strelka.sources={0},{1}\n"
 
-    ##TODO (cgates): deprecated; remove
-    @staticmethod
-    def validate_input_file(meta_headers, dummy_column_header):
-        return "##source=strelka" in meta_headers
-
     @staticmethod
     def _is_strelka_vcf(file_reader):
         if file_reader.file_name.endswith(".vcf"):

jacquard/variant_caller_transforms/variant_caller_factory.py

@@ -16,6 +16,7 @@
                              "Strelka": strelka.VERSION}
 
 class VariantCallerFactory(object):
+    #pylint: disable=too-few-public-methods
     def __init__(self, args=None):
         self._callers = [varscan.Varscan(args),
                          strelka.Strelka(),

jacquard/variant_caller_transforms/varscan.py

@@ -196,18 +196,6 @@ def _get_hc_file_pattern(args):
 
         return compiled_regex
 
-    ##TODO (cgates): deprecated; remove
-    @staticmethod
-    def validate_input_file(meta_headers, column_header):
-        if "##source=VarScan2" not in meta_headers:
-            return 0
-
-        if _VARSCAN_SOMATIC_HEADER == column_header:
-            return 1
-        else:
-            raise utils.JQException("Unexpected VarScan VCF structure - "
-                                    "missing NORMAL and TUMOR headers.")
-
     @staticmethod
     def _validate_filter_file(file_reader):
         column_header = 0
@@ -394,7 +382,7 @@ def _create_vcf_readers(pair_tuples):
 
 #pylint: disable=too-many-locals
     def claim(self, file_readers):
-        """Recognizes and claims MuTect VCFs form the set of all input VCFs.
+        """Recognizes and claims VarScan VCFs form the set of all input VCFs.
 
         Each defined caller has a chance to evaluate and claim all the incoming
         files as something that it can process. Since VarScan can claim
@@ -405,7 +393,7 @@ def claim(self, file_readers):
             file_readers: the collection of currently unclaimed files
 
         Returns:
-            A tuple of unclaimed readers and MuTectVcfReaders.
+            A tuple of unclaimed readers and VarScanVcfReaders.
         """
 
         (prefix_to_readers,

test/utils/test_case.py

@@ -74,17 +74,17 @@ def assertCommand(self, command, expected_dir):
 
         output = command[2]
 
-        #try:
-        if os.path.isfile(output):
-            output_file = os.path.basename(output)
-            self._compare_files(output, output_file, expected_dir)
-        elif os.path.isdir(output):
-            for output_file in os.listdir(output):
-                new_output = os.path.join(output, output_file)
-                self._compare_files(new_output, output_file, expected_dir)
-        #except self.failureException as e:
-        #    msg = "discrepancy in command [{}]: {}".format(" ".join(command), e)
-        #    raise self.failureException(msg)
+        try:
+            if os.path.isfile(output):
+                output_file = os.path.basename(output)
+                self._compare_files(output, output_file, expected_dir)
+            elif os.path.isdir(output):
+                for output_file in os.listdir(output):
+                    new_output = os.path.join(output, output_file)
+                    self._compare_files(new_output, output_file, expected_dir)
+        except self.failureException as e:
+            msg = "discrepancy in command [{}]: {}".format(" ".join(command), e)
+            raise self.failureException(msg)
 
     def entab(self, string, old="|"):
         return string.replace(old, "\t")

test/variant_caller_transforms/common_tags_test.py

@@ -32,14 +32,14 @@ def test_somatic_tag(self):
         self.assertEquals("Integer", tag.vcf_type)
         self.assertEquals("1", tag.vcf_number)
 
-    def test_add_tag_values_is_abstract(self):
-        try:
-            class FakeTag(common_tags.JacquardTag):
-                def __init__(self): pass
-            FakeTag()
-            self.fail(("Should not be able to instantiate JacquardTag without "
-                       "overrideing abstract methods."))
-        except TypeError: pass
+    def test_add_tag_values_raisesNotImplementedError(self):
+        class FakeTag(common_tags.JacquardTag):
+            def __init__(self): pass
+        tag = FakeTag()
+        self.assertRaises(NotImplementedError,
+                          tag.add_tag_values,
+                          VcfRecord("1", "42", "A", "C")
+                          )
 
     def test_metaheader(self):
         tag_type = common_tags.JacquardTag.GENOTYPE_TAG

test/variant_caller_transforms/mutect_test.py

@@ -129,14 +129,6 @@ def setUp(self):
         super(MutectTestCase, self).setUp()
         self.caller = mutect.Mutect()
 
-    def test_validateInputFile_isValid(self):
-        metaheaders = ["##MuTect=blah"]
-        self.assertTrue(self.caller.validate_input_file(metaheaders, "#column_header"))
-
-    def test_validateInputFile_isNotValid(self):
-        metaheaders = ["Foo"]
-        self.assertFalse(self.caller.validate_input_file(metaheaders, "#column_header"))
-
     def test_claim(self):
         record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
         content1 = ["##foo", "##source=strelka", "#chrom", record1]