def test_duplicate_pairs_no_discard(self):
# Two identical pairs. Ensure only one is emitted
p = test_utils.pair1()
# use the first read to create the map-reduce key
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p)) # add it twice
self.__reducer.discard_duplicates = False
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(
4, len(self.__ctx.emitted.values()
[0])) # four SAM records associated with the same key
flags = map(lambda sam: int(*re.match("(\d+).*", sam).groups(1)),
self.__ctx.emitted.values()[0])
# ensure we have two marked as duplicates
self.assertEqual(
2, len(filter(lambda flag: flag & sam_flags.SAM_FDP, flags)))
# ensure we have two NOT marked as duplicates
self.assertEqual(
2, len(filter(lambda flag: flag & sam_flags.SAM_FDP == 0, flags)))
# check counter
if self.__ctx.counters.has_key(self.__frag_counter_name()):
self.assertEqual(0,
self.__ctx.counters[self.__frag_counter_name()])
self.assertTrue(self.__ctx.counters.has_key(
self.__pair_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__pair_counter_name()])
def test_duplicate_fragments_read1_no_discard(self):
# load pair 1 and erase its second read
p = list(test_utils.pair1())
p = test_utils.erase_read2(p)
p0 = p[0]
# insert the pair into the context, twice
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__reducer.discard_duplicates = False
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(2,
len(self.__ctx.emitted.values()
[0])) # Two SAM records associated with the key
short_name = p0.get_name()[0:-2]
self.assertEqual(short_name, self.__ctx.emitted.keys()[0])
flags = map(lambda sam: int(*re.match("(\d+).*", sam).groups(1)),
self.__ctx.emitted.values()[0])
# ensure we have one marked as duplicate
self.assertEqual(
1, len(filter(lambda flag: flag & sam_flags.SAM_FDP, flags)))
# and ensure we have one NOT marked as duplicates
self.assertEqual(
1, len(filter(lambda flag: flag & sam_flags.SAM_FDP == 0, flags)))
# check counter
self.assertFalse(
self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(
self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_rmdup_bug(self):
test_case_data = [
"HWI-ST200R_251:5:1208:19924:124635#GCCAAT\t83\t20\t6181935\t60\t5S96M\t=\t6181919\t-112\tAAGTGGAAGATTTGGGAATCTGAGTGGATTTGGTAACAGTAGAGGGGTGGATCTGGCTTGGAAAACAATCGAGGTACCAATATAGGTGGTAGATGAATTTT\t?<?AADADBFBF<EHIGHGGGEAF3AF<CHGGDG9?GHFFACDHH)?@AHEHHIIIIE>A=A:?);B27@;@?>,;;C(5::>>>@5:()4>@@@######\tXC:i:96\tXT:A:U\tNM:i:1\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:1\tXO:i:0\tXG:i:0\tMD:Z:13G82\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:######@@@>4)(:5@>>>::5(C;;,>?@;@72B;)?:A=A>EIIIIHHEHA@?)HHDCAFFHG?9GDGGHC<FA3FAEGGGHGIHE<FBFBDADAA?<?",
"HWI-ST200R_251:5:1208:19924:124635#GCCAAT\t163\t20\t6181919\t60\t101M\t=\t6181935\t112\tCTGAGCACACCAAAATTCATCTACCACCTATATTGGTACCTCGATTGTTTTCCAAGCCAGATCCACACCTCTACTGTTACCAAATCCACTCAGATTCCCAA\t@@@FFFDDFHG??;EEH>HHGIGHEGCGEGGIGJG31?DDBBD>FGG@HG??DFBBADFAGII3@EH;;CEHECBB7?>CE.;...5>ACDDA:C:;>:>?\tXT:A:U\tNM:i:2\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:2\tXO:i:0\tXG:i:0\tMD:Z:29G36C34\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:@@@FFFDDFHG??;EEH>HHGIGHEGCGEGGIGJG31?DDBBD>FGG@HG??DFBBADFAGII3@EH;;CEHECBB7?>CE.;...5>ACDDA:C:;>:>?",
"HWI-ST200R_251:6:2207:18561:163438#GCCAAT\t83\t20\t6181938\t60\t8S93M\t=\t6181919\t-112\tAAAATTCATCTACCACCTATATTGGTACCTCGATTGTTTTCCAAGCCAGATCCACCCCTCTACTGTTACCAAATCCACTCAGATTCCCAAATCTTCCACTT\t@@@DDFDFHHHHHJJJEHGGHIHHAEGHJJIJJFGGHGIDIGIJJ?BBGGGIIIJJIJGFHGIJEC(=3?C;?B9?@C>CECECAA(;;3>C#########\tXC:i:93\tXT:A:U\tNM:i:2\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:2\tXO:i:0\tXG:i:0\tMD:Z:10G36C45\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:#########C>3;;(AACECEC>C@?9B?;C?3=(CEJIGHFGJIJJIIIGGGBB?JJIGIDIGHGGFJJIJJHGEAHHIHGGHEJJJHHHHHFDFDD@@@",
"HWI-ST200R_251:6:2207:18561:163438#GCCAAT\t163\t20\t6181919\t60\t101M\t=\t6181938\t112\tCTGAGCACACCAAAATTCATCTACCACCTATATTGGTACCTCGATTGTTTTCCAAGCCAGATCCACACCTCTACTGTTACCAAATCCACTCAGATTCCCAA\t@CCFFDDFHHHHHIJJJIIJJJIJJIIJGJIIIJII?DGHIGHDGHIIIJIJIJIIDCHGIJIIGGHIFEHHHHFFFFFDC.6.66;@CCCDCCDC>CCCA\tXT:A:U\tNM:i:2\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:2\tXO:i:0\tXG:i:0\tMD:Z:29G36C34\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:@CCFFDDFHHHHHIJJJIIJJJIJJIIJGJIIIJII?DGHIGHDGHIIIJIJIJIIDCHGIJIIGGHIFEHHHHFFFFFDC.6.66;@CCCDCCDC>CCCA",
]
self.__reducer.discard_duplicates = False
sams = map(SAMMapping, test_case_data)
left_key = "0020:000006181919:F"
pairs = ((sams[1], sams[0]), (sams[3], sams[2]))
# add the pairs to the context
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[0]))
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[1]))
self.__reducer.reduce(self.__ctx)
self.assertEqual(2, len(self.__ctx.emitted))
self.assertEqual([2, 2], map(len, self.__ctx.emitted.itervalues()))
key = "HWI-ST200R_251:6:2207:18561:163438#GCCAAT"
good_pair_sam = self.__ctx.emitted[key]
for read in good_pair_sam:
mapping = SAMMapping("\t".join((key, read)))
self.assertFalse(mapping.is_duplicate())
key = "HWI-ST200R_251:5:1208:19924:124635#GCCAAT"
dup_pair_sam = self.__ctx.emitted[key]
for read in dup_pair_sam:
mapping = SAMMapping("\t".join((key, read)))
self.assertTrue(mapping.is_duplicate())
def test_fragment_with_duplicate_in_pair_1_no_discard(self):
# Ensure the reducer catches a fragment duplicate of pair[0]
p = list(test_utils.pair1())
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
p = test_utils.erase_read2(p)
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__reducer.discard_duplicates = False
self.__reducer.reduce(self.__ctx)
# now ensure that both were emitted, but the fragment is marked as duplicate
self.__ensure_pair1_emitted()
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(3, len(self.__ctx.emitted.values()[0])) # 3 SAM records associated with the key (for the pair)
# make sure we have a read with the duplicate flag set
regexp = "(\d+)\s+.*"
flags = [ int(re.match(regexp, value).group(1)) for value in self.__ctx.emitted.values()[0] ]
dup_flags = [ flag for flag in flags if flag & sam_flags.SAM_FDP ]
self.assertEqual(1, len(dup_flags))
f = dup_flags[0]
self.assertTrue( f & sam_flags.SAM_FR1 > 0 ) # ensure the duplicate read is r1
self.assertTrue( f & sam_flags.SAM_FPD == 0 ) # ensure the duplicate read is unpaired
# check counter
self.assertFalse(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_rmdup_bug(self):
test_case_data = [
"HWI-ST200R_251:5:1208:19924:124635#GCCAAT\t83\t20\t6181935\t60\t5S96M\t=\t6181919\t-112\tAAGTGGAAGATTTGGGAATCTGAGTGGATTTGGTAACAGTAGAGGGGTGGATCTGGCTTGGAAAACAATCGAGGTACCAATATAGGTGGTAGATGAATTTT\t?<?AADADBFBF<EHIGHGGGEAF3AF<CHGGDG9?GHFFACDHH)?@AHEHHIIIIE>A=A:?);B27@;@?>,;;C(5::>>>@5:()4>@@@######\tXC:i:96\tXT:A:U\tNM:i:1\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:1\tXO:i:0\tXG:i:0\tMD:Z:13G82\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:######@@@>4)(:5@>>>::5(C;;,>?@;@72B;)?:A=A>EIIIIHHEHA@?)HHDCAFFHG?9GDGGHC<FA3FAEGGGHGIHE<FBFBDADAA?<?",
"HWI-ST200R_251:5:1208:19924:124635#GCCAAT\t163\t20\t6181919\t60\t101M\t=\t6181935\t112\tCTGAGCACACCAAAATTCATCTACCACCTATATTGGTACCTCGATTGTTTTCCAAGCCAGATCCACACCTCTACTGTTACCAAATCCACTCAGATTCCCAA\t@@@FFFDDFHG??;EEH>HHGIGHEGCGEGGIGJG31?DDBBD>FGG@HG??DFBBADFAGII3@EH;;CEHECBB7?>CE.;...5>ACDDA:C:;>:>?\tXT:A:U\tNM:i:2\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:2\tXO:i:0\tXG:i:0\tMD:Z:29G36C34\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:@@@FFFDDFHG??;EEH>HHGIGHEGCGEGGIGJG31?DDBBD>FGG@HG??DFBBADFAGII3@EH;;CEHECBB7?>CE.;...5>ACDDA:C:;>:>?",
"HWI-ST200R_251:6:2207:18561:163438#GCCAAT\t83\t20\t6181938\t60\t8S93M\t=\t6181919\t-112\tAAAATTCATCTACCACCTATATTGGTACCTCGATTGTTTTCCAAGCCAGATCCACCCCTCTACTGTTACCAAATCCACTCAGATTCCCAAATCTTCCACTT\t@@@DDFDFHHHHHJJJEHGGHIHHAEGHJJIJJFGGHGIDIGIJJ?BBGGGIIIJJIJGFHGIJEC(=3?C;?B9?@C>CECECAA(;;3>C#########\tXC:i:93\tXT:A:U\tNM:i:2\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:2\tXO:i:0\tXG:i:0\tMD:Z:10G36C45\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:#########C>3;;(AACECEC>C@?9B?;C?3=(CEJIGHFGJIJJIIIGGGBB?JJIGIDIGHGGFJJIJJHGEAHHIHGGHEJJJHHHHHFDFDD@@@",
"HWI-ST200R_251:6:2207:18561:163438#GCCAAT\t163\t20\t6181919\t60\t101M\t=\t6181938\t112\tCTGAGCACACCAAAATTCATCTACCACCTATATTGGTACCTCGATTGTTTTCCAAGCCAGATCCACACCTCTACTGTTACCAAATCCACTCAGATTCCCAA\t@CCFFDDFHHHHHIJJJIIJJJIJJIIJGJIIIJII?DGHIGHDGHIIIJIJIJIIDCHGIJIIGGHIFEHHHHFFFFFDC.6.66;@CCCDCCDC>CCCA\tXT:A:U\tNM:i:2\tSM:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tAM:i:37\tX0:i:1\tX1:i:0\tXM:i:2\tXO:i:0\tXG:i:0\tMD:Z:29G36C34\tRG:Z:raw_merged-1.2.3.4.5.6.7.8.bam\tOQ:Z:@CCFFDDFHHHHHIJJJIIJJJIJJIIJGJIIIJII?DGHIGHDGHIIIJIJIJIIDCHGIJIIGGHIFEHHHHFFFFFDC.6.66;@CCCDCCDC>CCCA",
]
self.__reducer.discard_duplicates = False
sams = map(SAMMapping, test_case_data)
left_key = "0020:000006181919:F"
pairs = ( (sams[1], sams[0]), (sams[3], sams[2]) )
# add the pairs to the context
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[0]))
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[1]))
self.__reducer.reduce(self.__ctx)
self.assertEqual(2, len(self.__ctx.emitted))
self.assertEqual([2,2], map(len, self.__ctx.emitted.itervalues()))
key = "HWI-ST200R_251:6:2207:18561:163438#GCCAAT"
good_pair_sam = self.__ctx.emitted[key]
for read in good_pair_sam:
mapping = SAMMapping("\t".join( (key,read) ))
self.assertFalse(mapping.is_duplicate())
key = "HWI-ST200R_251:5:1208:19924:124635#GCCAAT"
dup_pair_sam = self.__ctx.emitted[key]
for read in dup_pair_sam:
mapping = SAMMapping("\t".join( (key,read) ))
self.assertTrue(mapping.is_duplicate())
def test_rmdup_clipped_unpaired(self):
test_case_data = [
"HWI-ST200R_251:5:1208:19924:124635#GCCAAT\t82\t20\t6181935\t60\t5S96M\t*\t0\t0\tAAGTGGAAGATTTGGGAATCTGAGTGGATTTGGTAACAGTAGAGGGGTGGATCTGGCTTGGAAAACAATCGAGGTACCAATATAGGTGGTAGATGAATTTT\t?<?AADADBFBF<EHIGHGGGEAF3AF<CHGGDG9?GHFFACDHH)?@AHEHHIIIIE>A=A:?);B27@;@?>,;;C(5::>>>@5:()4>@@@######\tXC:i:96",
"HWI-ST200R_251:6:2207:18561:163438#GCCAAT\t82\t20\t6181930\t60\t101M\t*\t0\t0\tAAGTGGAAGATTTGGGAATCTGAGTGGATTTGGTAACAGTAGAGGGGTGGATCTGGCTTGGAAAACAATCGAGGTACCAATATAGGTGGTAGATGAATTTT\t@CCFFDDFHHHHHIJJJIIJJJIJJIIJGJIIIJII?DGHIGHDGHIIIJIJIJIIDCHGIJIIGGHIFEHHHHFFFFFDC.6.66;@CCCDCCDC>CCCA",
]
self.__reducer.discard_duplicates = False
sams = map(SAMMapping, test_case_data)
left_key = "0020:000006181930:R"
pairs = ( (sams[0], None), (sams[1], None) )
# add the pairs to the context
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[0]))
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[1]))
self.__reducer.reduce(self.__ctx)
self.assertEqual(2, len(self.__ctx.emitted))
self.assertEqual([1,1], map(len, self.__ctx.emitted.itervalues()))
key = "HWI-ST200R_251:6:2207:18561:163438#GCCAAT"
good_read_sam = self.__ctx.emitted[key][0]
mapping = SAMMapping("\t".join( (key,good_read_sam) ))
self.assertFalse(mapping.is_duplicate())
key = "HWI-ST200R_251:5:1208:19924:124635#GCCAAT"
dup_read_sam = self.__ctx.emitted[key][0]
mapping = SAMMapping("\t".join( (key,dup_read_sam) ))
self.assertTrue(mapping.is_duplicate())
def test_rmdup_clipped_unpaired(self):
test_case_data = [
"HWI-ST200R_251:5:1208:19924:124635#GCCAAT\t82\t20\t6181935\t60\t5S96M\t*\t0\t0\tAAGTGGAAGATTTGGGAATCTGAGTGGATTTGGTAACAGTAGAGGGGTGGATCTGGCTTGGAAAACAATCGAGGTACCAATATAGGTGGTAGATGAATTTT\t?<?AADADBFBF<EHIGHGGGEAF3AF<CHGGDG9?GHFFACDHH)?@AHEHHIIIIE>A=A:?);B27@;@?>,;;C(5::>>>@5:()4>@@@######\tXC:i:96",
"HWI-ST200R_251:6:2207:18561:163438#GCCAAT\t82\t20\t6181930\t60\t101M\t*\t0\t0\tAAGTGGAAGATTTGGGAATCTGAGTGGATTTGGTAACAGTAGAGGGGTGGATCTGGCTTGGAAAACAATCGAGGTACCAATATAGGTGGTAGATGAATTTT\t@CCFFDDFHHHHHIJJJIIJJJIJJIIJGJIIIJII?DGHIGHDGHIIIJIJIJIIDCHGIJIIGGHIFEHHHHFFFFFDC.6.66;@CCCDCCDC>CCCA",
]
self.__reducer.discard_duplicates = False
sams = map(SAMMapping, test_case_data)
left_key = "0020:000006181930:R"
pairs = ((sams[0], None), (sams[1], None))
# add the pairs to the context
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[0]))
self.__ctx.add_value(left_key, proto.serialize_pair(pairs[1]))
self.__reducer.reduce(self.__ctx)
self.assertEqual(2, len(self.__ctx.emitted))
self.assertEqual([1, 1], map(len, self.__ctx.emitted.itervalues()))
key = "HWI-ST200R_251:6:2207:18561:163438#GCCAAT"
good_read_sam = self.__ctx.emitted[key][0]
mapping = SAMMapping("\t".join((key, good_read_sam)))
self.assertFalse(mapping.is_duplicate())
key = "HWI-ST200R_251:5:1208:19924:124635#GCCAAT"
dup_read_sam = self.__ctx.emitted[key][0]
mapping = SAMMapping("\t".join((key, dup_read_sam)))
self.assertTrue(mapping.is_duplicate())
def test_duplicate_fragments_read1(self):
# load pair 1
p = list(test_utils.pair1())
p = test_utils.erase_read2(p)
p0 = p[0]
# insert the pair into the context, twice
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(
1, len(self.__ctx.emitted.values()
[0])) # only one SAM record associated with the key
short_name = p0.get_name()[0:-2]
self.assertEqual(short_name, self.__ctx.emitted.keys()[0])
self.assertTrue(
re.match("\d+\s+%s\s+%d\s+.*" % (p0.tid, p0.pos),
self.__ctx.emitted[short_name][0]))
# check counter
self.assertFalse(
self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(
self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_fragment_with_duplicate_in_pair_1(self):
# Ensure the reducer catches a fragment duplicate of pair[0]
p = list(test_utils.pair1())
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
test_utils.erase_read2(p)
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
# now ensure that the pair was emitted, but not the fragment
self.__ensure_only_pair1_emitted()
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(2, len(self.__ctx.emitted.values()[0])) # two SAM records associated with the key (for the pair)
# check counter
self.assertFalse(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_duplicate_pairs(self):
# Two identical pairs. Ensure only one is emitted
p = test_utils.pair1()
# use the first read to create the map-reduce key
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p)) # add it twice
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(2, len(self.__ctx.emitted.values()[0])) # two SAM records associated with the same key
self.__ensure_only_pair1_emitted()
# check counter
if self.__ctx.counters.has_key(self.__frag_counter_name()):
self.assertEqual(0, self.__ctx.counters[self.__frag_counter_name()])
self.assertTrue(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__pair_counter_name()])
def process(self, pair):
if any(pair):
# order pair such that left-most read is at pos 0
ordered_pair = self.__order_pair(pair)
record = protobuf_mapping.serialize_pair(ordered_pair)
# emit with the left coord
key = self.get_hit_key(ordered_pair[0])
self.ctx.emit(key, record)
if ordered_pair[0].is_mapped():
self.event_monitor.count("mapped coordinates", 1)
# since we ordered the pair, if ordered_pair[0] is unmapped
# ordered_pair[1] will not be mapped.
if ordered_pair[1]:
if ordered_pair[1].is_mapped():
# a full pair. We emit the coordinate, but with PAIR_STRING as the value
key = self.get_hit_key(ordered_pair[1])
self.ctx.emit(key, seqal_app.PAIR_STRING)
self.event_monitor.count("mapped coordinates", 1)
else:
self.event_monitor.count("unmapped reads", 1)
else:
self.event_monitor.count("unmapped reads", len(pair))
# in all cases, forward the original pair to the link in the chain
if self.next_link:
self.next_link.process(pair)
def process(self, pair):
if any(pair):
# order pair such that left-most read is at pos 0
ordered_pair = self.__order_pair(pair)
record = protobuf_mapping.serialize_pair(ordered_pair)
# emit with the left coord
key = self.get_hit_key(ordered_pair[0])
self.ctx.emit(key, record)
if ordered_pair[0].is_mapped():
self.event_monitor.count("mapped coordinates", 1)
# since we ordered the pair, if ordered_pair[0] is unmapped
# ordered_pair[1] will not be mapped.
if ordered_pair[1]:
if ordered_pair[1].is_mapped():
# a full pair. We emit the coordinate, but with PAIR_STRING as the value
key = self.get_hit_key(ordered_pair[1])
self.ctx.emit(key, seqal_app.PAIR_STRING)
self.event_monitor.count("mapped coordinates", 1)
else:
self.event_monitor.count("unmapped reads", 1)
else:
self.event_monitor.count("unmapped reads", len(pair))
# in all cases, forward the original pair to the link in the chain
if self.next_link:
self.next_link.process(pair)
def test_unmapped2(self):
p = test_utils.pair1()
p[1].set_mapped(False)
p[0].set_mate_mapped(False)
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
self.assertEqual(2, len(self.__ctx.emitted.values()[0]))
def test_emit_on_left_key(self):
# load pair 1
p = test_utils.pair1()
# use the first read to create the map-reduce key
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
self.__ensure_only_pair1_emitted()
def test_unmapped2(self):
p = test_utils.pair1()
p[1].set_mapped(False)
p[0].set_mate_mapped(False)
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
self.assertEqual(2, len(self.__ctx.emitted.values()[0]))
def test_fw_rev_missed_dup_pair(self):
"""
Here we have two duplicate pairs, as detected by Picard.
The reads 10364 have flags 'pPR1', 'pPr2'.
The reads 138222 have flags 'pPR2d', 'pPr1d'.
10364/1 has no trimming and is on the fw strand. It should have a key position 88888399:F.
10364/2 is also on the rev strand. We expect a key position of 88888421+53 = 88888474:R.
138222/1 is on the rev strand. We expect a key position of 88888404+70 = 88888474:R.
138222/2 has been trimmed but is on the fw strand. It should have a key position 88888399:F.
"""
test_case_data = [
"HWI-ST332_97:3:66:16214:10364#0\t99\t10\t88888399\t46\t76M\t=\t88888421\t76\tTGATTTTGCTCCATTGTCTTCTAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCATTCTGGAAGCTTATCTATTG\tHGHHHHHHFHEHHHHGHHFHHHGGHHHHHHHHHHEHDHGEHFHHHHHHHGGHHHHHHHHHHBHBBEGG=GFFFF@F",
"HWI-ST332_97:3:66:16214:10364#0\t147\t10\t88888421\t46\t22S54M\t=\t88888399\t-76\tTGATTCCGCTCCATGTGCCTCGAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCATTCTGGAAGCTTATCTATTG\t#######################AA@EHGHEHHHHHHHHHHHHHHHEHHHHHHFHFHHHHHHHHHHHHHHHGHHFH",
"HWI-ST332_97:3:7:10556:138222#0\t1107\t10\t88888404\t23\t5S71M\t=\t88888399\t-76\tTGATGTTGCTCCATTGTCTTCTAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCATTCTGGAAGCTTATCTATTG\t######C<BCC:B,DDDC=BD3@CB8B?DBD@E@EEEEECED@CDB=8C7A@D=DEDEDCDBECDE<>;;,17,45",
"HWI-ST332_97:3:7:10556:138222#0\t1187\t10\t88888399\t15\t50M26S\t=\t88888404\t76\tTGATTTTGCTCCATTGTCTTCTAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCCGTCTGGTTGCTTCTATTTTG\tCC7EEFFF@FHHFHHG?GGF:>4.7GD8DC@D>CCFGG@GGG5GG4<CB###########################",
]
sams = map(SAMMapping, test_case_data)
pair1 = sams[0:2]
pair2 = sams[2:]
left_key = "0010:000088888399:F"
self.__ctx.add_value(left_key, proto.serialize_pair(pair1))
self.__ctx.add_value(left_key, proto.serialize_pair(pair2))
self.__reducer.reduce(self.__ctx)
# verify emitted data. 2 keys (one per read name) and 2 reads for each key
self.assertEqual(2, len(self.__ctx.emitted))
self.assertEqual([2, 2], map(len, self.__ctx.emitted.itervalues()))
key = "HWI-ST332_97:3:66:16214:10364#0"
good_pair_sam = self.__ctx.emitted[key]
for read in good_pair_sam:
mapping = SAMMapping("\t".join((key, read)))
self.assertFalse(mapping.is_duplicate())
key = "HWI-ST332_97:3:7:10556:138222#0"
dup_pair_sam = self.__ctx.emitted[key]
for read in dup_pair_sam:
mapping = SAMMapping("\t".join((key, read)))
self.assertTrue(mapping.is_duplicate())
self.setUp() # clean-up and repeat for the right key
right_key = "0010:000088888474:R"
self.__ctx.add_value(right_key, PAIR_STRING)
self.__ctx.add_value(right_key, PAIR_STRING)
self.__reducer.reduce(self.__ctx)
# verify no data emitted
self.assertEqual(0, len(self.__ctx.emitted))
def test_fw_rev_missed_dup_pair(self):
"""
Here we have two duplicate pairs, as detected by Picard.
The reads 10364 have flags 'pPR1', 'pPr2'.
The reads 138222 have flags 'pPR2d', 'pPr1d'.
10364/1 has no trimming and is on the fw strand. It should have a key position 88888399:F.
10364/2 is also on the rev strand. We expect a key position of 88888421+53 = 88888474:R.
138222/1 is on the rev strand. We expect a key position of 88888404+70 = 88888474:R.
138222/2 has been trimmed but is on the fw strand. It should have a key position 88888399:F.
"""
test_case_data = [
"HWI-ST332_97:3:66:16214:10364#0\t99\t10\t88888399\t46\t76M\t=\t88888421\t76\tTGATTTTGCTCCATTGTCTTCTAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCATTCTGGAAGCTTATCTATTG\tHGHHHHHHFHEHHHHGHHFHHHGGHHHHHHHHHHEHDHGEHFHHHHHHHGGHHHHHHHHHHBHBBEGG=GFFFF@F",
"HWI-ST332_97:3:66:16214:10364#0\t147\t10\t88888421\t46\t22S54M\t=\t88888399\t-76\tTGATTCCGCTCCATGTGCCTCGAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCATTCTGGAAGCTTATCTATTG\t#######################AA@EHGHEHHHHHHHHHHHHHHHEHHHHHHFHFHHHHHHHHHHHHHHHGHHFH",
"HWI-ST332_97:3:7:10556:138222#0\t1107\t10\t88888404\t23\t5S71M\t=\t88888399\t-76\tTGATGTTGCTCCATTGTCTTCTAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCATTCTGGAAGCTTATCTATTG\t######C<BCC:B,DDDC=BD3@CB8B?DBD@E@EEEEECED@CDB=8C7A@D=DEDEDCDBECDE<>;;,17,45",
"HWI-ST332_97:3:7:10556:138222#0\t1187\t10\t88888399\t15\t50M26S\t=\t88888404\t76\tTGATTTTGCTCCATTGTCTTCTAGCTTGTGTTATGCCTGTTGAAAGTACAAAATCCGTCTGGTTGCTTCTATTTTG\tCC7EEFFF@FHHFHHG?GGF:>4.7GD8DC@D>CCFGG@GGG5GG4<CB###########################",
]
sams = map(SAMMapping, test_case_data)
pair1 = sams[0:2]
pair2 = sams[2:]
left_key = "0010:000088888399:F"
self.__ctx.add_value(left_key, proto.serialize_pair(pair1))
self.__ctx.add_value(left_key, proto.serialize_pair(pair2))
self.__reducer.reduce(self.__ctx)
# verify emitted data. 2 keys (one per read name) and 2 reads for each key
self.assertEqual(2, len(self.__ctx.emitted))
self.assertEqual([2,2], map(len, self.__ctx.emitted.itervalues()))
key = "HWI-ST332_97:3:66:16214:10364#0"
good_pair_sam = self.__ctx.emitted[key]
for read in good_pair_sam:
mapping = SAMMapping("\t".join( (key,read) ))
self.assertFalse(mapping.is_duplicate())
key = "HWI-ST332_97:3:7:10556:138222#0"
dup_pair_sam = self.__ctx.emitted[key]
for read in dup_pair_sam:
mapping = SAMMapping("\t".join( (key,read) ))
self.assertTrue(mapping.is_duplicate())
self.setUp() # clean-up and repeat for the right key
right_key = "0010:000088888474:R"
self.__ctx.add_value(right_key, PAIR_STRING)
self.__ctx.add_value(right_key, PAIR_STRING)
self.__reducer.reduce(self.__ctx)
# verify no data emitted
self.assertEqual(0, len(self.__ctx.emitted))
def test_emit_on_left_key(self):
# load pair 1
p = test_utils.pair1()
# use the first read to create the map-reduce key
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
self.__ensure_only_pair1_emitted()
def test_unmapped1(self):
p = test_utils.pair1()
p[0].set_mapped(False)
p[1].set_mate_mapped(False)
# Having an unmapped read before a mapped read is not allowed. This should
# raise an exception
# The key is meaningless
self.__ctx.add_value(test_utils.make_key(p[1]), proto.serialize_pair(p))
self.assertRaises(ValueError, self.__reducer.reduce, self.__ctx)
def test_duplicate_fragments_read1(self):
# load pair 1
p = list(test_utils.pair1())
p = test_utils.erase_read2(p)
p0 = p[0]
# insert the pair into the context, twice
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(1, len(self.__ctx.emitted.values()[0])) # only one SAM record associated with the key
short_name = p0.get_name()[0:-2]
self.assertEqual(short_name, self.__ctx.emitted.keys()[0])
self.assertTrue( re.match("\d+\s+%s\s+%d\s+.*" % (p0.tid, p0.pos), self.__ctx.emitted[short_name][0]) )
# check counter
self.assertFalse(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_unmapped1(self):
p = test_utils.pair1()
p[0].set_mapped(False)
p[1].set_mate_mapped(False)
# Having an unmapped read before a mapped read is not allowed. This should
# raise an exception
# The key is meaningless
self.__ctx.add_value(test_utils.make_key(p[1]),
proto.serialize_pair(p))
self.assertRaises(ValueError, self.__reducer.reduce, self.__ctx)
def test_duplicate_pairs_no_discard(self):
# Two identical pairs. Ensure only one is emitted
p = test_utils.pair1()
# use the first read to create the map-reduce key
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p)) # add it twice
self.__reducer.discard_duplicates = False
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(4, len(self.__ctx.emitted.values()[0])) # four SAM records associated with the same key
flags = map(lambda sam: int(*re.match("(\d+).*", sam).groups(1)), self.__ctx.emitted.values()[0])
# ensure we have two marked as duplicates
self.assertEqual(2, len(filter(lambda flag: flag & sam_flags.SAM_FDP, flags)) )
# ensure we have two NOT marked as duplicates
self.assertEqual(2, len(filter(lambda flag: flag & sam_flags.SAM_FDP == 0, flags)) )
# check counter
if self.__ctx.counters.has_key(self.__frag_counter_name()):
self.assertEqual(0, self.__ctx.counters[self.__frag_counter_name()])
self.assertTrue(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__pair_counter_name()])
def test_fragment_with_duplicate_in_pair_1(self):
# Ensure the reducer catches a fragment duplicate of pair[0]
p = list(test_utils.pair1())
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
test_utils.erase_read2(p)
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__reducer.reduce(self.__ctx)
# now ensure that the pair was emitted, but not the fragment
self.__ensure_only_pair1_emitted()
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(2, len(
self.__ctx.emitted.values()
[0])) # two SAM records associated with the key (for the pair)
# check counter
self.assertFalse(
self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(
self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_duplicate_pairs(self):
# Two identical pairs. Ensure only one is emitted
p = test_utils.pair1()
# use the first read to create the map-reduce key
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p)) # add it twice
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(
2, len(self.__ctx.emitted.values()
[0])) # two SAM records associated with the same key
self.__ensure_only_pair1_emitted()
# check counter
if self.__ctx.counters.has_key(self.__frag_counter_name()):
self.assertEqual(0,
self.__ctx.counters[self.__frag_counter_name()])
self.assertTrue(self.__ctx.counters.has_key(
self.__pair_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__pair_counter_name()])
def test_fragment_with_duplicate_in_pair_1_no_discard(self):
# Ensure the reducer catches a fragment duplicate of pair[0]
p = list(test_utils.pair1())
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
p = test_utils.erase_read2(p)
self.__ctx.add_value(test_utils.make_key(p[0]),
proto.serialize_pair(p))
self.__reducer.discard_duplicates = False
self.__reducer.reduce(self.__ctx)
# now ensure that both were emitted, but the fragment is marked as duplicate
self.__ensure_pair1_emitted()
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(3, len(
self.__ctx.emitted.values()
[0])) # 3 SAM records associated with the key (for the pair)
# make sure we have a read with the duplicate flag set
regexp = "(\d+)\s+.*"
flags = [
int(re.match(regexp, value).group(1))
for value in self.__ctx.emitted.values()[0]
]
dup_flags = [flag for flag in flags if flag & sam_flags.SAM_FDP]
self.assertEqual(1, len(dup_flags))
f = dup_flags[0]
self.assertTrue(
f & sam_flags.SAM_FR1 > 0) # ensure the duplicate read is r1
self.assertTrue(
f
& sam_flags.SAM_FPD == 0) # ensure the duplicate read is unpaired
# check counter
self.assertFalse(
self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(
self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_duplicate_fragments_read1_no_discard(self):
# load pair 1 and erase its second read
p = list(test_utils.pair1())
p = test_utils.erase_read2(p)
p0 = p[0]
# insert the pair into the context, twice
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__ctx.add_value(test_utils.make_key(p[0]), proto.serialize_pair(p))
self.__reducer.discard_duplicates = False
self.__reducer.reduce(self.__ctx)
self.assertEqual(1, len(self.__ctx.emitted.keys()))
self.assertEqual(2, len(self.__ctx.emitted.values()[0])) # Two SAM records associated with the key
short_name = p0.get_name()[0:-2]
self.assertEqual(short_name, self.__ctx.emitted.keys()[0])
flags = map(lambda sam: int(*re.match("(\d+).*", sam).groups(1)), self.__ctx.emitted.values()[0])
# ensure we have one marked as duplicate
self.assertEqual(1, len(filter(lambda flag: flag & sam_flags.SAM_FDP, flags)) )
# and ensure we have one NOT marked as duplicates
self.assertEqual(1, len(filter(lambda flag: flag & sam_flags.SAM_FDP == 0, flags)) )
# check counter
self.assertFalse(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_fragment_with_duplicate_in_pair_2(self):
# Ensure the reducer catches a fragment duplicate of pair[1].
p = list(test_utils.pair1())
# Insert the pair into the context
self.__ctx.add_value(test_utils.make_key(p[1]), PAIR_STRING)
# Remove the first read from the pair, reorder so that the None is at index 1,
# the serialize and insert into the context.
test_utils.erase_read1(p)
self.__ctx.add_value(test_utils.make_key(p[1]), proto.serialize_pair( (p[1], None) ))
self.__reducer.reduce(self.__ctx)
# now ensure that nothing was emitted. The pair isn't emitted because
# the key refers to read2, and the fragment isn't emitted because it's a duplicate of
# the one in the pair.
self.assertEqual(0, len(self.__ctx.emitted.keys()))
# check counter
self.assertFalse(self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def test_fragment_with_duplicate_in_pair_2(self):
# Ensure the reducer catches a fragment duplicate of pair[1].
p = list(test_utils.pair1())
# Insert the pair into the context
self.__ctx.add_value(test_utils.make_key(p[1]), PAIR_STRING)
# Remove the first read from the pair, reorder so that the None is at index 1,
# the serialize and insert into the context.
test_utils.erase_read1(p)
self.__ctx.add_value(test_utils.make_key(p[1]),
proto.serialize_pair((p[1], None)))
self.__reducer.reduce(self.__ctx)
# now ensure that nothing was emitted. The pair isn't emitted because
# the key refers to read2, and the fragment isn't emitted because it's a duplicate of
# the one in the pair.
self.assertEqual(0, len(self.__ctx.emitted.keys()))
# check counter
self.assertFalse(
self.__ctx.counters.has_key(self.__pair_counter_name()))
self.assertTrue(self.__ctx.counters.has_key(
self.__frag_counter_name()))
self.assertEqual(1, self.__ctx.counters[self.__frag_counter_name()])
def __pipe_pair_through(self, pair):
message = io.serialize_pair(pair)
return io.unserialize_pair(message)
def __pipe_pair_through(self, pair):
message = io.serialize_pair(pair)
return io.unserialize_pair(message)
def test_empty_read1(self):
# Ensure the reducer raises an exception if the pair[0] is None
p = test_utils.erase_read1(list(test_utils.pair1()))
self.__ctx.add_value(test_utils.make_key(p[1]),
proto.serialize_pair(p))
self.assertRaises(ValueError, self.__reducer.reduce, self.__ctx)
def test_empty_read1(self):
# Ensure the reducer raises an exception if the pair[0] is None
p = test_utils.erase_read1(list(test_utils.pair1()))
self.__ctx.add_value(test_utils.make_key(p[1]), proto.serialize_pair(p))
self.assertRaises(ValueError, self.__reducer.reduce, self.__ctx)
