Merge remote-tracking branch 'origin/hotfix'

docs/conf.py

@@ -69,7 +69,7 @@
 # The short X.Y version.
 version = '1.5'
 # The full version, including alpha/beta/rc tags.
-release = '1.5.6'
+release = '1.5.7'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/index.rst

@@ -52,12 +52,10 @@ New in release |release|
 
 This is a hotfix release. It fixes the following issues:
 
-- The ``pvacbind run`` command would previously error out if one of the input
-  sequences would contain a X stop codon. This update will remove the X amino
-  acid and the downstream sequence before further processing the remaining
-  protein sequence.
-- A bug in the ``pvacfuse top_score_filter`` code would previsouly result
-  in an error when trying to run this command. This has now been fixed.
+- The ``pvacbind run`` command would previously allow fasta input files with
+  duplicated headers. However, it would silently skip subsequent entries with
+  duplicated headers even if the fasta sequence was novel. With this release
+  pVACbind will now error out if a duplicate fasta header is encounterd.
 
 New in version |version|
 ------------------------

docs/releases/1_5.rst

@@ -156,9 +156,21 @@ This is a hotfix release. It fixes the following issues:
 1.5.6
 -----
 
+This is a hotfix release. It fixes the following issues:
+
 - The ``pvacbind run`` command would previously error out if one of the input
   sequences would contain a X stop codon. This update will remove the X amino
   acid and the downstream sequence before further processing the remaining
   protein sequence.
 - A bug in the ``pvacfuse top_score_filter`` code would previsouly result
   in an error when trying to run this command. This has now been fixed.
+
+1.5.7
+-----
+
+This is a hotfix release. It fixes the following issues:
+
+- The ``pvacbind run`` command would previously allow fasta input files with
+  duplicated headers. However, it would silently skip subsequent entries with
+  duplicated headers even if the fasta sequence was novel. With this release
+  pVACbind will now error out if a duplicate fasta header is encounterd.

lib/pipeline.py

@@ -491,8 +491,12 @@ def split_fasta_basename(self, length):
 
     def uniquify_records(self, records):
         fasta_sequences = OrderedDict()
+        ids = []
         for record in records:
+            if record.id in ids:
+                raise Exception("Duplicate fasta header {}. Please ensure that the input FASTA uses unique headers.".format(record.id))
             fasta_sequences.setdefault(str(record.seq), []).append(record.id)
+            ids.append(record.id)
         count = 1
         uniq_records = []
         keys = {}

lib/top_score_filter.py

@@ -79,9 +79,9 @@ def execute(self):
                             filtered_lines.append(line_with_max_expression)
                         else:
                             line_with_lowest_transcript_id = lines[0]
-                            lowest_transcript_id = re.compile('ENST(\d+)').match(line_with_lowest_transcript_id['Transcript']).group(1)
+                            lowest_transcript_id = re.compile(r'ENST(\d+)').match(line_with_lowest_transcript_id['Transcript']).group(1)
                             for line in lines:
-                                transcript_id = re.compile('ENST(\d+)').match(line['Transcript']).group(1)
+                                transcript_id = re.compile(r'ENST(\d+)').match(line['Transcript']).group(1)
                                 if transcript_id < lowest_transcript_id:
                                     lowest_transcript_id = transcript_id
                                     line_with_lowest_transcript_id = line

setup.py

@@ -61,7 +61,7 @@
 
 setup(
     name="pvactools",
-    version="1.5.6",
+    version="1.5.7",
     packages=[
         "tools",
         "tools.pvacbind",

tests/test_data/pvacbind/input.duplicate_header.fasta

@@ -0,0 +1,4 @@
+>1
+LPLPPPPLLPLLLLLLGASGG
+>1
+LPLPPPPLLPLLPLLLLLGASGG

tests/test_pvacbind.py

@@ -273,23 +273,39 @@ def test_pvacbind_pipeline(self):
 
             with self.assertRaises(SystemExit) as cm:
                 run.main([
-                    os.path.join(self.test_data_directory, "input.vcf"),
+                    os.path.join(self.test_data_directory, "input.fasta"),
                     'Test',
                     'H2-IAb',
                     'NNalign',
                     output_dir.name,
-                    '--top-score-metric=lowest',
                     '--keep-tmp-files',
                 ])
-                self.assertEqual(
-                    cm.exception,
-                    "Restart inputs are different from past inputs: \n" +
-                    "Past input: downstream_sequence_length - None\n" +
-                    "Current input: downstream_sequence_length - 1000"
-                )
+            self.assertEqual(
+                str(cm.exception),
+                "Restart inputs are different from past inputs: \n" +
+                "Past input: top_score_metric - lowest\n" +
+                "Current input: top_score_metric - median\nAborting."
+            )
 
             output_dir.cleanup()
 
+    def test_duplicate_fasta_header(self):
+        with self.assertRaises(Exception) as cm:
+            output_dir = tempfile.TemporaryDirectory(dir = self.test_data_directory)
+            run.main([
+                os.path.join(self.test_data_directory, "input.duplicate_header.fasta"),
+                'Test',
+                'HLA-A*02:01',
+                'NetMHC',
+                output_dir.name,
+                '-e', '8'
+            ])
+        self.assertEqual(
+            str(cm.exception),
+            "Duplicate fasta header 1. Please ensure that the input FASTA uses unique headers."
+        )
+        output_dir.cleanup()
+
     def test_pvacbind_combine_and_condense_steps(self):
         output_dir = tempfile.TemporaryDirectory(dir = self.test_data_directory)
         for subdir in ['MHC_Class_I', 'MHC_Class_II']:

tests/test_pvacseq.py

@@ -422,12 +422,12 @@ def test_pvacseq_pipeline(self):
                     '--top-score-metric=lowest',
                     '--keep-tmp-files',
                 ])
-                self.assertEqual(
-                    cm.exception,
-                    "Restart inputs are different from past inputs: \n" +
-                    "Past input: downstream_sequence_length - None\n" +
-                    "Current input: downstream_sequence_length - 1000"
-                )
+            self.assertEqual(
+                str(cm.exception),
+                "Restart inputs are different from past inputs: \n" +
+                "Past input: downstream_sequence_length - None\n" +
+                "Current input: downstream_sequence_length - 1000\nAborting."
+            )
 
             output_dir.cleanup()