Merge pull request #17 from qiyunzhu/bugfix

Bug fix for duplicated sample name check

plate_mapper/plate_mapper.py

@@ -164,43 +164,43 @@ def plate_mapper(input_f, barseq_f, output_f, names_f=None, special_f=None,
     output_f.close()
     print('  Done.')
 
+    # Check for repeated sample names
+    warning = ''
+    samples = Counter(samples)
+    repeated = [name for name, count in samples.items() if count > 1]
+    if repeated:
+        warning += ('  Repeated samples: %s.\n'
+                    % _print_list(sorted(repeated)))
+
     # Validate sample names
     if names_f:
         print('Validating sample names...')
-        samples = Counter(samples)
         names = set()
         for line in names_f:
             l = line.rstrip().split('\t')
             if l[0]:  # skip empty names
                 names.add(l[0])  # keep first field as name
         names_f.close()
-        warning = ''
         if names:
             sample_set = set(samples)
             # samples in plate map but not in name list
             novel = sample_set - names
             # samples in name list but not in plate map
             missing = names - sample_set
-            # samples that occur more than one times in plate map
-            repeated = set()
-            for name in names:
-                if name in samples and samples[name] > 1:
-                    repeated.add(name)
             if novel:
                 warning += ('  Novel samples: %s.\n'
                             % _print_list(sorted(novel)))
             if missing:
                 warning += ('  Missing samples: %s.\n'
                             % _print_list(sorted(missing)))
-            if repeated:
-                warning += ('  Repeated samples: %s.\n'
-                            % _print_list(sorted(repeated)))
         print('  Done.')
-        if warning:
-            # display warning message
-            warnings.formatwarning = lambda msg, *a: str(msg)
-            warnings.warn('Warning:\n%s' % warning)
-        print('Task completed.')
+
+    # Display warning message
+    if warning:
+        warnings.formatwarning = lambda msg, *a: str(msg)
+        warnings.warn('Warning:\n%s' % warning)
+
+    print('Task completed.')
 
 
 if __name__ == "__main__":

plate_mapper/tests/data/exp_mapping.txt

@@ -25,7 +25,7 @@ sp220	TATGCCAGAGAT	ATCG	1	C4	QZ	8/15/16
 sp012	CCTCGCATGACC	ATCG	3	A1	QZ	10/26/16
 sp015	GGCGTAACGGCA	ATCG	3	A2	QZ	10/26/16
 sp018	GCGAGGAAGTCC	ATCG	3	A3	QZ	10/26/16
-blank4A	CAAATTCGGGAT	ATCG	3	A4	QZ	10/26/16
+sp020	CAAATTCGGGAT	ATCG	3	A4	QZ	10/26/16
 sp013	CGCGCAAGTATT	ATCG	3	B1	QZ	10/26/16
 sp016	AATACAGACCTG	ATCG	3	B2	QZ	10/26/16
 sp019	GGACAAGTGCGA	ATCG	3	B3	QZ	10/26/16

plate_mapper/tests/data/plate_map.txt

@@ -4,6 +4,6 @@ B	sp002	sp005	sp007	missing4B		a sample is missing
 C	sp003	blank2C	blank3C	sp220
 
 Plate#2	1	2	3	4	#	Who  When
-A	sp012	sp015	sp018	blank4A	3	QZ	10/26/16
+A	sp012	sp015	sp018	sp020	3	QZ	10/26/16
 B	sp013	sp016	sp019
 C

plate_mapper/tests/data/plate_map_w_dup.txt

@@ -0,0 +1,9 @@
+Plate#1	1	2	3	4	#	Who  When
+A	sp001	sp004	sp006	blank4A	1	QZ	8/15/16
+B	sp002	sp005	sp007	missing4B		a sample is missing
+C	sp003	blank2C	blank3C	sp220
+
+Plate#2	1	2	3	4	#	Who  When
+A	sp001	sp015	sp018	sp020	3	QZ	10/26/16
+B	sp013	sp016	sp019
+C

plate_mapper/tests/data/sample_list.txt

@@ -13,6 +13,7 @@ sp016
 sp017
 sp018
 sp019
+sp020
 blank2C
 blank3C
 blank4A

plate_mapper/tests/test_plate_mapper.py

@@ -110,6 +110,18 @@ def test_plate_mapper(self):
         err = 'Error: Code \'+\' has no name.'
         self.assertEqual(str(context.exception), err)
 
+        # test a successful conversion with repeated sample name warning
+        input_f = open(join(datadir, 'plate_map_w_dup.txt'), 'r')
+        barseq_f = open(join(datadir, 'barseq_temp.txt'), 'r')
+        output_f = open(obs_output_fp, 'w')
+        # check screen warning message
+        msg = ('Warning:\n'
+               '  Repeated samples: sp001.\n')
+        with catch_warnings(record=True) as w:
+            simplefilter('always')
+            plate_mapper(input_f, barseq_f, output_f)
+            assert msg in str(w[-1].message)
+
         # test a successful conversion with sample name validation warnings
         input_f = open(join(datadir, 'plate_map.txt'), 'r')
         barseq_f = open(join(datadir, 'barseq_temp.txt'), 'r')
@@ -119,8 +131,7 @@ def test_plate_mapper(self):
         # check screen warning message
         msg = ('Warning:\n'
                '  Novel samples: missing4B, sp220.\n'
-               '  Missing samples: sp014, sp017.\n'
-               '  Repeated samples: blank4A.\n')
+               '  Missing samples: sp014, sp017.\n')
         with catch_warnings(record=True) as w:
             simplefilter('always')
             plate_mapper(input_f, barseq_f, output_f, names_f)