From b303c1be196d90538f7e8b8ca4c28e47b76c99d8 Mon Sep 17 00:00:00 2001
From: Chris Tomkins-Tinch <tomkinsc@gmail.com>
Date: Thu, 28 Feb 2019 20:55:32 -0500
Subject: [PATCH] allow outlier_barcodes to act on single-index runs (#932)

* initial changes to allow outlier_barcodes to act on single-index runs

* cruft removal
---
 .../single_index/barcodes.txt                 | 1000 +++++++++++++++++
 .../single_index/expected.txt                 |    5 +
 .../single_index/metrics.txt                  |   36 +
 test/unit/test_illumina.py                    |   14 +
 util/illumina_indices.py                      |   69 +-
 5 files changed, 1096 insertions(+), 28 deletions(-)
 create mode 100644 test/input/TestIlluminaBarcodeHelper/single_index/barcodes.txt
 create mode 100644 test/input/TestIlluminaBarcodeHelper/single_index/expected.txt
 create mode 100644 test/input/TestIlluminaBarcodeHelper/single_index/metrics.txt

diff --git a/test/input/TestIlluminaBarcodeHelper/single_index/barcodes.txt b/test/input/TestIlluminaBarcodeHelper/single_index/barcodes.txt
new file mode 100644
index 000000000..5487c0bed
--- /dev/null
+++ b/test/input/TestIlluminaBarcodeHelper/single_index/barcodes.txt
@@ -0,0 +1,1000 @@
+Barcode1	Likely_Index_Names1	Barcode2	Likely_Index_Names2	Count
+CAGAGAGG	N708		Unknown	2236892
+GGACTCCT	N705		Unknown	1481969
+TAGGCATG	N706		Unknown	1388954
+CAGGCGAT	Unknown		Unknown	1375668
+GCTACGCT	N709		Unknown	1370424
+CTCTCTAC	N707		Unknown	1181498
+GTAGAGGA	N712		Unknown	1144504
+CGAGGCTG	N710		Unknown	1078586
+AAGAGGCA	N711		Unknown	910195
+TACAGCAT	Unknown		Unknown	670800
+CATTTTAT	Unknown		Unknown	627525
+TCATTCAT	Unknown		Unknown	624071
+ACTGATAT	Unknown		Unknown	613337
+ATTCCTAT	Unknown		Unknown	578494
+CGGAATAT	Unknown		Unknown	564153
+GACGACAT	Unknown		Unknown	555371
+TATAATAT	Unknown		Unknown	545928
+CATGGCAT	Unknown		Unknown	455796
+ATGAGCAT	Unknown		Unknown	449580
+CCAACAAT	Unknown		Unknown	446825
+TAATCGAT	Unknown		Unknown	409191
+CAACTAAT	Unknown		Unknown	402513
+TCCCGAAT	Unknown		Unknown	340067
+CAAAAGAT	Unknown		Unknown	305712
+CTAGCTAT	Unknown		Unknown	298921
+CACCGGAT	Unknown		Unknown	217902
+GAAGAGGA	N712		Unknown	8740
+AGAGAGGA	Unknown		Unknown	8486
+GCTACGAT	N709		Unknown	6491
+GGACTCAT	N705		Unknown	5693
+AAGAGAGG	N708		Unknown	5249
+CAGGCATG	N706		Unknown	5051
+CAGGGAGG	N708		Unknown	4417
+CAGAGAGA	N708,S511		Unknown	4238
+CAGAGAGT	N708		Unknown	4217
+GTAGAGGT	N712		Unknown	4012
+TAATTCAT	Unknown		Unknown	4012
+GATACGCT	N709		Unknown	3956
+CTTCCTAT	Unknown		Unknown	3741
+GGAGAGGA	N712		Unknown	3650
+CTCTCTAA	E502,N707,S502,[N|S|E]502		Unknown	3421
+CAGGCGAG	Unknown		Unknown	3409
+AAGAGGAA	N711		Unknown	3380
+CAGACGAT	Unknown		Unknown	3260
+GTAGAGAA	N712		Unknown	3247
+CAGAGAAG	N708		Unknown	3185
+ACTGATCT	Unknown		Unknown	3119
+CATTTATC	Unknown		Unknown	3101
+TATAATCT	Unknown		Unknown	2948
+GCAACGCT	N709		Unknown	2912
+CGAGGCTA	N710		Unknown	2893
+TAGAGAGG	N708		Unknown	2785
+CAGAGCGG	N708		Unknown	2785
+GGACTCTA	Unknown		Unknown	2749
+CAGAGGGG	N708		Unknown	2693
+TAGGCAAG	N706		Unknown	2675
+CAAGAGGA	Unknown		Unknown	2615
+ATCTCTAC	N707		Unknown	2608
+CGAGGCTT	N710		Unknown	2566
+AAGGCGAT	S520		Unknown	2561
+CTACGCTA	Unknown		Unknown	2553
+TAGGCATT	N706		Unknown	2541
+GGACTACT	N705		Unknown	2499
+CCAACATC	Unknown		Unknown	2479
+GGAATCCT	N705		Unknown	2477
+CAGGAGAT	Unknown		Unknown	2476
+TATAAATC	Unknown		Unknown	2454
+GAGAGAGG	N708		Unknown	2441
+CAAGCGAT	Unknown		Unknown	2437
+TATATATC	Unknown		Unknown	2436
+CTCTCTAT	E502,S502,[N|S|E]502		Unknown	2419
+CAGGCGCT	Unknown		Unknown	2404
+ATTTTATC	Unknown		Unknown	2402
+GCACGCTA	Unknown		Unknown	2320
+CAGAGGCA	N711		Unknown	2304
+CAAAGATC	Unknown		Unknown	2268
+GCTACCTA	Unknown		Unknown	2250
+CAGAGGGA	Unknown		Unknown	2217
+TAGGCGAT	Unknown		Unknown	2190
+CGAGGCAG	N710		Unknown	2168
+TCATCATC	Unknown		Unknown	2148
+GAGGCTGA	Unknown		Unknown	2128
+GAGGCATG	N706		Unknown	2123
+CTCTCACA	Unknown		Unknown	2080
+GGGCTCCT	N705		Unknown	2028
+GCTCGCTA	N716		Unknown	2027
+CGGAATCT	Unknown		Unknown	2024
+CTATCTAC	N707		Unknown	2014
+TAGCATGA	Unknown		Unknown	2003
+GCTAGCTA	Unknown		Unknown	1985
+TAAATATC	Unknown		Unknown	1983
+GCTACGTA	Unknown		Unknown	1964
+CCTCTCTT	Unknown		Unknown	1964
+AGAGGCTG	N710		Unknown	1937
+CAGGCGAA	Unknown		Unknown	1922
+CAGAGGAT	Unknown		Unknown	1888
+CTCTCTCC	N707		Unknown	1872
+TCCCGATC	Unknown		Unknown	1863
+CAGCGAGG	N708		Unknown	1857
+CAGAGAAT	Unknown		Unknown	1837
+CAGAAGGA	Unknown		Unknown	1786
+TAGGAATG	N706		Unknown	1757
+AATGATAT	Unknown		Unknown	1745
+CGAGGATG	N710		Unknown	1740
+CTCCTACA	Unknown		Unknown	1725
+CAACAATC	Unknown		Unknown	1723
+GCAGAGGA	N712		Unknown	1702
+TAAGCATG	N706		Unknown	1689
+TTCTCTAC	N707		Unknown	1664
+GCGACGCT	N709		Unknown	1664
+GTACGCTA	Unknown		Unknown	1663
+GTAGAGGC	D502,N712		Unknown	1646
+CAGGCAAT	Unknown		Unknown	1641
+CAGGGGAT	Unknown		Unknown	1625
+AAGAGGCT	N711		Unknown	1612
+CAAACAAT	Unknown		Unknown	1596
+AGGCGATC	Unknown		Unknown	1596
+CAGGCATC	Unknown		Unknown	1590
+CCTGATAT	Unknown		Unknown	1582
+TAGACATG	N706		Unknown	1572
+GCTAAGCT	N709		Unknown	1567
+GAACTCCT	N705		Unknown	1567
+CAACTATC	Unknown		Unknown	1565
+CGAGGCGA	Unknown		Unknown	1553
+AGGCATGA	Unknown		Unknown	1526
+CCTTCTCT	Unknown		Unknown	1522
+CGAGCTGA	Unknown		Unknown	1519
+CATGCGAT	Unknown		Unknown	1508
+TGACTCCT	N705		Unknown	1498
+CAGGAGGA	Unknown		Unknown	1498
+GTCTCTAC	N707		Unknown	1487
+TACAGCAG	Unknown		Unknown	1474
+CTCTTACA	Unknown		Unknown	1469
+AAGGCATG	N706		Unknown	1463
+GTAGCGGA	N712		Unknown	1442
+CGAGAGGA	Unknown		Unknown	1433
+GAGGCGAT	Unknown		Unknown	1421
+TCTCTACA	Unknown		Unknown	1413
+CCTACGCT	N709		Unknown	1402
+TACAGAAT	Unknown		Unknown	1395
+CATAATAT	Unknown		Unknown	1373
+TCTACGCT	E504,N709,[N|S|E]504		Unknown	1364
+CGACTCCT	N705		Unknown	1364
+ATGGCATC	Unknown		Unknown	1352
+TCTCCTCT	E503,E504,S503,[N|S|E]503,[N|S|E]504		Unknown	1346
+GGAGGCTG	N710		Unknown	1335
+CTGAGCAT	Unknown		Unknown	1325
+CTCTCTCA	Unknown		Unknown	1312
+CAAAGAGG	N708		Unknown	1302
+GGTCTCCT	N705		Unknown	1292
+GGCTCCTA	Unknown		Unknown	1285
+CGGGATAT	Unknown		Unknown	1284
+GTACTCCT	N705		Unknown	1278
+CTCACTAC	N707		Unknown	1278
+CAGCGATC	Unknown		Unknown	1275
+TAGGCATA	N706		Unknown	1273
+TAAGCATC	Unknown		Unknown	1272
+CGGAGAGG	N708,S511		Unknown	1266
+AGGAATAT	Unknown		Unknown	1261
+TAGGCGTG	N706		Unknown	1259
+CAAGGCAT	Unknown		Unknown	1257
+ACTGAATC	Unknown		Unknown	1250
+CAGGCTAT	S520		Unknown	1241
+CAGGCAAG	Unknown		Unknown	1241
+CAGACAGG	N708		Unknown	1234
+GGACCCTA	Unknown		Unknown	1234
+TACCGAAT	Unknown		Unknown	1233
+AATTTTAT	Unknown		Unknown	1232
+TCTTCCTT	Unknown		Unknown	1225
+CACTCTAC	N707		Unknown	1223
+CGGAAATC	Unknown		Unknown	1222
+CACAGCAT	Unknown		Unknown	1220
+AAAAGGCA	N711		Unknown	1214
+ACTGCTAT	Unknown		Unknown	1209
+CATAGCAT	Unknown		Unknown	1205
+ACTGAGAT	Unknown		Unknown	1203
+CTTCTACA	Unknown		Unknown	1200
+GAAGACAT	Unknown		Unknown	1199
+GGACTGCT	N705		Unknown	1198
+GACTCCTA	Unknown		Unknown	1197
+TAAAGCAT	Unknown		Unknown	1196
+TAGGGATG	N706		Unknown	1187
+CCAAAATC	Unknown		Unknown	1168
+TGAGGCTG	N710		Unknown	1160
+GGACTCCA	N705		Unknown	1148
+ATTACTAT	Unknown		Unknown	1147
+CGGATATC	Unknown		Unknown	1135
+GGATCCTA	Unknown		Unknown	1133
+GTAGAGGG	N712		Unknown	1123
+GCTACACT	N709		Unknown	1122
+AGAGGCAA	Unknown		Unknown	1122
+TAGGCAGA	Unknown		Unknown	1120
+CAGTCGAT	Unknown		Unknown	1120
+CGAGGGTG	N710		Unknown	1119
+AAGAGGCC	N711		Unknown	1109
+ATTCATAT	Unknown		Unknown	1101
+GGACCCCT	N705		Unknown	1100
+TCCCGCAT	Unknown		Unknown	1095
+CATTCTAT	Unknown		Unknown	1092
+AAGACGCA	N711		Unknown	1086
+CAGAATAT	Unknown		Unknown	1070
+AAGAGCAA	Unknown		Unknown	1068
+AAGAGACA	N711		Unknown	1065
+GTAAGGAA	E505,S505,[N|S|E]505		Unknown	1062
+CCACAATC	Unknown		Unknown	1061
+CAGGGCAT	Unknown		Unknown	1057
+TACAGCAA	Unknown		Unknown	1052
+CAGGTGAT	Unknown		Unknown	1037
+TCCCTCCT	Unknown		Unknown	1034
+TACACATC	Unknown		Unknown	1033
+ACTGACAT	Unknown		Unknown	1032
+TAGAGGAA	Unknown		Unknown	1030
+CTCGCTAC	N707		Unknown	1029
+GCTACTCT	E504,N709,[N|S|E]504		Unknown	1027
+GGAGTCCT	N705		Unknown	1025
+GGCCTCCT	N705		Unknown	1022
+GCTACGCA	N709		Unknown	1014
+CAATCGAT	Unknown		Unknown	1009
+CAGAGATG	N708		Unknown	1002
+CAAGGCTG	N710		Unknown	999
+CGGGGCTG	N710		Unknown	999
+TAAACGAT	Unknown		Unknown	989
+TATACTAT	Unknown		Unknown	987
+TCACGAAT	Unknown		Unknown	977
+CAAAAATC	Unknown		Unknown	977
+GCTGCGCT	N709		Unknown	976
+TATCGATC	Unknown		Unknown	971
+GGAATATC	Unknown		Unknown	969
+CCGGCGAT	Unknown		Unknown	964
+AAGGGGCA	N711		Unknown	956
+TTATTCAT	Unknown		Unknown	956
+CACGACAT	Unknown		Unknown	950
+TAGGCCTG	N706		Unknown	949
+TATAAGAT	Unknown		Unknown	943
+CTTCCTCT	Unknown		Unknown	942
+GACAACAT	Unknown		Unknown	942
+CAAGAGAT	Unknown		Unknown	941
+TCATTAAT	Unknown		Unknown	937
+TATGATAT	Unknown		Unknown	932
+CGGGCTGA	Unknown		Unknown	932
+AAGAGCAT	Unknown		Unknown	931
+GACGACAG	Unknown		Unknown	928
+GAGAGGAA	Unknown		Unknown	927
+CATATTAT	Unknown		Unknown	926
+GGCGACAT	Unknown		Unknown	925
+TAGGCTGA	Unknown		Unknown	924
+CGAGCCTG	N710		Unknown	923
+CAAACGAT	Unknown		Unknown	922
+CCGAGAGG	N708		Unknown	918
+CAGGGATG	Unknown		Unknown	917
+TCATTCAG	Unknown		Unknown	907
+TAGGCATC	N706		Unknown	903
+TACGACAT	Unknown		Unknown	900
+CTCTTCTT	Unknown		Unknown	900
+TAATCGAG	Unknown		Unknown	898
+CCTTTTAT	Unknown		Unknown	897
+TATGCATG	N706		Unknown	896
+GGACACCT	N705		Unknown	896
+ATGAGAAT	Unknown		Unknown	885
+CGGAAGAT	Unknown		Unknown	878
+CAGAGTGG	N708		Unknown	875
+ATTCCTCT	Unknown		Unknown	872
+GACGACAA	Unknown		Unknown	872
+GACGAAAT	Unknown		Unknown	871
+CATTTCAT	Unknown		Unknown	869
+TACAACAT	Unknown		Unknown	869
+TTCCTCCT	Unknown		Unknown	869
+CGAGGCAT	Unknown		Unknown	861
+CTCTATAC	N707		Unknown	860
+CAGGCCAT	Unknown		Unknown	859
+ATGAGCAG	N722		Unknown	858
+CGGACTAT	Unknown		Unknown	854
+TATAACAT	Unknown		Unknown	849
+TTTCCTAT	Unknown		Unknown	849
+ACTACGCT	N709		Unknown	848
+AATGGCAT	Unknown		Unknown	848
+TCAATCAT	Unknown		Unknown	845
+GACGCCAT	Unknown		Unknown	843
+TGGCATGA	Unknown		Unknown	843
+GAGACATC	Unknown		Unknown	842
+CTACTATC	Unknown		Unknown	836
+GCTACGCG	N709		Unknown	833
+CATGGCAG	Unknown		Unknown	832
+TAATAGAT	Unknown		Unknown	831
+ACAACAAT	Unknown		Unknown	830
+TAACGATC	Unknown		Unknown	825
+ATGATATC	Unknown		Unknown	825
+CAGAAGAT	Unknown		Unknown	816
+AAACTAAT	Unknown		Unknown	815
+TAGGCAGG	N706		Unknown	813
+GACGAATC	Unknown		Unknown	813
+CTGGCGAT	Unknown		Unknown	812
+CCAGGCGA	Unknown		Unknown	811
+CAGAGCAT	Unknown		Unknown	808
+TAGGCAAT	Unknown		Unknown	802
+TATAAAAT	Unknown		Unknown	800
+GATACGAT	Unknown		Unknown	798
+CATTTTAG	Unknown		Unknown	798
+TATTTTAT	Unknown		Unknown	797
+CGGGCGAT	Unknown		Unknown	792
+GCTAGGCT	N709		Unknown	790
+CGAAATAT	Unknown		Unknown	789
+TGATTCAT	Unknown		Unknown	788
+TACACCAT	Unknown		Unknown	788
+GTTCCTAT	Unknown		Unknown	786
+TCTAATAT	Unknown		Unknown	785
+TGGAATAT	Unknown		Unknown	782
+CAACTCAT	Unknown		Unknown	774
+TCCTTCCT	Unknown		Unknown	773
+GGACGCCT	N705		Unknown	773
+TGGGCATG	N706		Unknown	770
+CCTCTACA	Unknown		Unknown	770
+CCTTTTCT	Unknown		Unknown	765
+CAGCCGAT	Unknown		Unknown	763
+AAAAGATC	Unknown		Unknown	761
+TACAGCCT	D501		Unknown	761
+CAAATAAT	Unknown		Unknown	759
+CAGGCGGT	Unknown		Unknown	758
+ATTCCGAT	Unknown		Unknown	757
+TACGGCAT	Unknown		Unknown	755
+TCATTCAA	Unknown		Unknown	754
+CTGAGAGG	N708		Unknown	753
+TCAACAAT	Unknown		Unknown	753
+GCTACCCT	N709		Unknown	752
+GCCACGCT	N709		Unknown	750
+ACTGAAAT	Unknown		Unknown	749
+CATTTGAT	Unknown		Unknown	747
+CTCTCAAC	N707		Unknown	743
+GCTATGCT	N709		Unknown	737
+AACTAATC	Unknown		Unknown	736
+GCTACTAT	Unknown		Unknown	736
+CTGTCTAC	N707		Unknown	736
+GTCGACAT	Unknown		Unknown	735
+TAATCGCT	Unknown		Unknown	734
+TAGGATGA	Unknown		Unknown	732
+AGACTCCT	N705		Unknown	731
+TATAATAG	Unknown		Unknown	730
+CATTTAAT	Unknown		Unknown	726
+CATGCCAT	Unknown		Unknown	726
+CAGAAAGG	N708		Unknown	726
+TATAGCAT	D501		Unknown	723
+TATGGCAT	Unknown		Unknown	722
+GAGACGCT	Unknown		Unknown	721
+AAGAGGAT	Unknown		Unknown	718
+GGACTCCG	N705		Unknown	718
+ATGAGCAA	Unknown		Unknown	717
+TCGGCATG	N706		Unknown	716
+GACGAGAT	Unknown		Unknown	716
+GTAGAGAT	Unknown		Unknown	716
+AAGAGGGA	N711		Unknown	713
+CGGAAAAT	Unknown		Unknown	711
+CCATTCAT	Unknown		Unknown	707
+ACCGGATC	Unknown		Unknown	704
+TAGCTATC	Unknown		Unknown	703
+ATCCTATC	Unknown		Unknown	701
+CATGGAAT	Unknown		Unknown	701
+TAAAAGAT	Unknown		Unknown	700
+ACTAATAT	Unknown		Unknown	697
+GCCGACAT	Unknown		Unknown	692
+TAGAGCAT	Unknown		Unknown	691
+CATGACAT	Unknown		Unknown	690
+TACAGGAT	Unknown		Unknown	688
+ACTGATAG	Unknown		Unknown	688
+CGAGGTGA	Unknown		Unknown	687
+TGTAATAT	Unknown		Unknown	685
+CGTTTTAT	Unknown		Unknown	685
+CCCCCCCC	Unknown		Unknown	684
+GCTCCGCT	N709		Unknown	684
+TAAGCGAT	Unknown		Unknown	678
+CGAAGCTG	N710		Unknown	676
+CGGCATAT	Unknown		Unknown	673
+TATAGTAT	Unknown		Unknown	669
+GGTACGCT	N709		Unknown	666
+GACGACCT	Unknown		Unknown	664
+AAAAAAAA	Unknown		Unknown	664
+CAGGGGGG	Unknown		Unknown	661
+CAGGCGTT	Unknown		Unknown	659
+TCATTCCT	Unknown		Unknown	659
+CGGAATAG	Unknown		Unknown	658
+CAGTGAGG	N708		Unknown	657
+CATAGAGG	N708		Unknown	656
+GACAGCAT	Unknown		Unknown	655
+GTAAAGGA	N712		Unknown	653
+TATAATAA	Unknown		Unknown	652
+AAAAAGAT	Unknown		Unknown	651
+ATAAGCAT	Unknown		Unknown	651
+TAACTAAT	Unknown		Unknown	651
+CTAGCGAT	Unknown		Unknown	650
+CGGAACAT	Unknown		Unknown	649
+GTTACGCT	N709		Unknown	647
+GAAGAGAT	Unknown		Unknown	646
+TAATCGAA	Unknown		Unknown	645
+CAACTGAT	Unknown		Unknown	645
+AAGGAGAT	Unknown		Unknown	642
+CCAACGAT	Unknown		Unknown	640
+ATTCCCAT	Unknown		Unknown	635
+ACAGATAT	Unknown		Unknown	635
+GAAGAGAA	Unknown		Unknown	635
+GGACTTCT	N705		Unknown	634
+TACATCAT	Unknown		Unknown	634
+CAGGCGAC	Unknown		Unknown	632
+ATTCCTAG	Unknown		Unknown	632
+TCAGTCAT	Unknown		Unknown	628
+AAGAAGCA	N711		Unknown	625
+GTAGGGAA	Unknown		Unknown	624
+CAATTTAT	Unknown		Unknown	623
+AATCCTAT	Unknown		Unknown	623
+CAGGCAGG	Unknown		Unknown	623
+CATGTTAT	Unknown		Unknown	621
+CATTATAT	Unknown		Unknown	621
+AATAATAT	Unknown		Unknown	620
+TAATTGAT	Unknown		Unknown	620
+CGAGGCGG	N710		Unknown	620
+CCCCCCCT	Unknown		Unknown	618
+TAAAATAT	Unknown		Unknown	617
+TAGGTATG	N706		Unknown	611
+CTCTCGAC	N707		Unknown	609
+CACGGATC	Unknown		Unknown	604
+GTCCTCCT	Unknown		Unknown	604
+TTTAATAT	Unknown		Unknown	603
+ATTCCTAA	Unknown		Unknown	598
+CGCGGCTG	N710		Unknown	595
+CTCTCCAC	N707		Unknown	595
+GGACTCTT	N705		Unknown	593
+TAGGCACG	N706		Unknown	591
+CACGGCAT	Unknown		Unknown	590
+CAGATAGG	N708		Unknown	590
+GAGAGGCA	N711		Unknown	589
+ATTCCAAT	Unknown		Unknown	589
+CAAAAAAT	Unknown		Unknown	589
+GCTCCTCT	Unknown		Unknown	589
+CATTTTCT	Unknown		Unknown	588
+GATGACAT	Unknown		Unknown	588
+CCACTAAT	Unknown		Unknown	585
+CACTAATC	Unknown		Unknown	585
+AATAGGCA	N711		Unknown	585
+AAGAGGCG	N711		Unknown	585
+GCACTCCT	N705		Unknown	584
+CCCCTCCT	Unknown		Unknown	583
+TTCTCTTC	Unknown		Unknown	581
+GGACTCCC	N705		Unknown	581
+CAGGGATC	Unknown		Unknown	580
+TCGTTCAT	Unknown		Unknown	578
+CCAACAAG	Unknown		Unknown	577
+TTGGCATG	N706		Unknown	577
+CTATCTAT	E502,S502,[N|S|E]502		Unknown	576
+CGTAATAT	Unknown		Unknown	574
+CCAACACT	Unknown		Unknown	573
+AAGGGCAA	Unknown		Unknown	570
+GGGAATAT	Unknown		Unknown	569
+ACTGATAA	Unknown		Unknown	569
+GCTACGCC	N709		Unknown	568
+GAAGAGGT	Unknown		Unknown	567
+CATTTTAA	Unknown		Unknown	566
+TCTTTCAT	Unknown		Unknown	561
+TCACTCAT	Unknown		Unknown	561
+TAGTCATG	N706		Unknown	559
+ATGGGCAT	Unknown		Unknown	557
+GTAGATGA	N712		Unknown	557
+CTTTCTAC	N707		Unknown	556
+AGGAGGCA	N711		Unknown	551
+GCATTCAT	Unknown		Unknown	550
+ATTCGTAT	Unknown		Unknown	550
+CGAATATC	Unknown		Unknown	549
+CAACTACT	Unknown		Unknown	549
+ATAGCTAT	Unknown		Unknown	547
+CAGGCGTG	Unknown		Unknown	543
+CTAACTAT	Unknown		Unknown	543
+CTTTTTAT	Unknown		Unknown	542
+TCCAGAAT	Unknown		Unknown	541
+TTAGAGGA	N712		Unknown	541
+TCATCGAT	Unknown		Unknown	540
+CAACTAAG	Unknown		Unknown	538
+CATGCTAT	Unknown		Unknown	537
+AAGAGCCA	N711		Unknown	536
+CAAAAGAG	Unknown		Unknown	536
+GTGAGGAA	Unknown		Unknown	534
+ATTGATAT	Unknown		Unknown	533
+CATGGCAA	Unknown		Unknown	533
+TACCGCAT	Unknown		Unknown	533
+TAGTCGAT	Unknown		Unknown	531
+CGAGACTG	N710		Unknown	531
+CAGGGAAG	Unknown		Unknown	531
+CTCCCTAC	E505,N707,S505,[N|S|E]505		Unknown	531
+CGACTAAT	S510		Unknown	530
+TCCCGACT	Unknown		Unknown	529
+ACGAGGCA	N711		Unknown	526
+CAGAGACG	N708		Unknown	525
+TCCTTCAT	Unknown		Unknown	524
+CGGCGATC	Unknown		Unknown	524
+ATACCTAT	Unknown		Unknown	523
+CCGCGCGG	Unknown		Unknown	523
+CATGATAT	Unknown		Unknown	521
+CCAACCAT	Unknown		Unknown	521
+CACAGGAT	Unknown		Unknown	520
+CAATAATC	Unknown		Unknown	519
+CTAGATAT	Unknown		Unknown	516
+CAGACAAT	Unknown		Unknown	515
+ATTGCTAT	Unknown		Unknown	515
+CCATTTTA	Unknown		Unknown	511
+CAAGCTAT	Unknown		Unknown	509
+AACGACAT	Unknown		Unknown	508
+TCCCGGAT	Unknown		Unknown	506
+ATGAGCCT	Unknown		Unknown	505
+TTCAGCAT	Unknown		Unknown	504
+CAACCAAT	Unknown		Unknown	502
+CCAAAAAT	Unknown		Unknown	500
+AACAGCAT	Unknown		Unknown	500
+TACAGATC	Unknown		Unknown	499
+GGATTCCT	N705		Unknown	499
+CGCTCTAC	N707		Unknown	497
+GATAAGAT	Unknown		Unknown	496
+GTAGGGGA	N712		Unknown	493
+TCCCGAAG	Unknown		Unknown	493
+CATGCATC	Unknown		Unknown	490
+CTAACAAT	Unknown		Unknown	490
+TCCAGCAT	Unknown		Unknown	486
+AAGCGGCA	N711		Unknown	485
+CTAGCTAG	Unknown		Unknown	485
+CATGGTAT	Unknown		Unknown	484
+ACTGGTAT	Unknown		Unknown	482
+TCATTGAT	Unknown		Unknown	481
+ATGACATC	Unknown		Unknown	480
+GCTATCTC	Unknown		Unknown	479
+ATTAGCAT	Unknown		Unknown	477
+CAAAGAAT	Unknown		Unknown	476
+ATTCTATC	Unknown		Unknown	475
+CGAGGCCG	N710		Unknown	475
+GGAGAGGT	Unknown		Unknown	474
+TATCATAT	Unknown		Unknown	472
+GACGGCAT	Unknown		Unknown	472
+TGCAGCAT	Unknown		Unknown	472
+TACCGGAT	Unknown		Unknown	471
+GATAATAT	Unknown		Unknown	471
+CTAGGCTG	N710		Unknown	470
+CTAGCAAT	Unknown		Unknown	468
+CATCTTAT	Unknown		Unknown	467
+CCAGCAAT	Unknown		Unknown	467
+CAACGGAT	Unknown		Unknown	465
+CAAAAGCT	Unknown		Unknown	465
+CACGCGAT	Unknown		Unknown	464
+ATGAGGAT	Unknown		Unknown	464
+CATGTCAT	Unknown		Unknown	463
+GTGGAGGA	N712		Unknown	463
+CCCCGAAT	Unknown		Unknown	461
+TAATCAAT	Unknown		Unknown	459
+GAAAAGAT	Unknown		Unknown	457
+TACAGTAT	Unknown		Unknown	456
+AAGATGCA	N711		Unknown	456
+CAGGCGGG	Unknown		Unknown	456
+GTAGAAGA	N712		Unknown	455
+AAAGGCAA	Unknown		Unknown	455
+ATTCTTAT	Unknown		Unknown	454
+CATACGAT	Unknown		Unknown	454
+CCCAACAA	Unknown		Unknown	453
+ATGAGGCA	N711		Unknown	452
+CGAGTCTG	N710		Unknown	451
+CTCTGTAC	N707		Unknown	449
+CCTCTTCC	Unknown		Unknown	449
+ATTCCCTA	Unknown		Unknown	447
+GCAACAAT	Unknown		Unknown	447
+CTAGCCAT	Unknown		Unknown	444
+CAGGGGAG	Unknown		Unknown	443
+CACCGAAT	Unknown		Unknown	442
+ATGACCAT	Unknown		Unknown	438
+AACCGGAT	Unknown		Unknown	437
+CAGAGCGT	Unknown		Unknown	435
+CTGGCTAT	Unknown		Unknown	434
+AGAGCATC	Unknown		Unknown	434
+CCTACGAT	Unknown		Unknown	434
+TCTGATAT	Unknown		Unknown	434
+GAGGCGCT	Unknown		Unknown	430
+TCCCGAAA	Unknown		Unknown	429
+CACCGGAG	Unknown		Unknown	429
+GCTACGTT	N709		Unknown	429
+CATTGTAT	Unknown		Unknown	429
+CGGAATAA	Unknown		Unknown	428
+CTCTACAT	Unknown		Unknown	428
+CAGACGCT	Unknown		Unknown	427
+TAATTAAT	Unknown		Unknown	424
+CCAACAAA	Unknown		Unknown	420
+ATAGAGGA	D502,N712		Unknown	420
+AAGAGTCA	N711		Unknown	417
+GTAGACGA	N712		Unknown	414
+CATGGGAT	Unknown		Unknown	414
+TCATTATC	Unknown		Unknown	413
+CGGAGTAT	Unknown		Unknown	412
+TAGAATAT	Unknown		Unknown	412
+CAAAGGAT	Unknown		Unknown	412
+GAAAAGGA	Unknown		Unknown	411
+GCAGAGGT	Unknown		Unknown	411
+CTACTAAT	Unknown		Unknown	411
+GGACTCGT	N705		Unknown	411
+CAGGGCGG	Unknown		Unknown	409
+GAACTAAT	Unknown		Unknown	408
+TAATCCAT	Unknown		Unknown	408
+CGTGGCTG	N710		Unknown	407
+CCTTTCCT	D503		Unknown	406
+TCCGAATC	Unknown		Unknown	405
+CGACTCAT	Unknown		Unknown	404
+CACCGGCT	Unknown		Unknown	402
+TCTCGAAT	Unknown		Unknown	401
+GAAGAGGG	Unknown		Unknown	398
+CTCTCTGC	N707		Unknown	397
+GCTGATAT	Unknown		Unknown	396
+CAGACAAG	Unknown		Unknown	396
+GATGGCAT	Unknown		Unknown	395
+CGGTATAT	Unknown		Unknown	394
+GTATAGGA	N712		Unknown	394
+CAACTAAA	Unknown		Unknown	393
+CCTGGCAT	Unknown		Unknown	393
+CATACGCT	Unknown		Unknown	392
+CTGAATAT	Unknown		Unknown	391
+GATTTTAT	Unknown		Unknown	390
+CATTTTTA	Unknown		Unknown	390
+TACGCATG	N706		Unknown	387
+TCATCCAT	Unknown		Unknown	387
+TCCTCTCT	Unknown		Unknown	387
+TAATGGAT	Unknown		Unknown	387
+CAATTCAT	Unknown		Unknown	386
+CTTTCTAT	E502,S502,[N|S|E]502		Unknown	386
+CAGCTAAT	Unknown		Unknown	386
+GCTACGGT	N709		Unknown	385
+CATGGCCT	Unknown		Unknown	383
+CAGGATCT	Unknown		Unknown	383
+GAGGACAT	Unknown		Unknown	382
+AAGAGAGA	Unknown		Unknown	382
+GAGGAGAT	Unknown		Unknown	382
+CAAAAGAA	Unknown		Unknown	381
+ATGAACAT	Unknown		Unknown	380
+CCAAGAAT	Unknown		Unknown	379
+TCATTTAT	Unknown		Unknown	379
+GAGCTCCT	Unknown		Unknown	378
+CCAATAAT	Unknown		Unknown	377
+TAGAGAAG	Unknown		Unknown	375
+GACGATAT	Unknown		Unknown	375
+CGTGGCAT	Unknown		Unknown	375
+GAGGAGGA	Unknown		Unknown	375
+CCAACTAA	Unknown		Unknown	374
+GTAGAGTA	N712		Unknown	371
+CTCTTTAC	E505,N707,S505,[N|S|E]505		Unknown	371
+GCAACGAT	Unknown		Unknown	369
+ATGAGTAT	Unknown		Unknown	368
+ACAGCATC	Unknown		Unknown	366
+CAGAGGGT	Unknown		Unknown	366
+CAAAATAT	Unknown		Unknown	365
+TAAAGAAT	Unknown		Unknown	365
+CAGGCTGA	Unknown		Unknown	363
+GAACTCAT	Unknown		Unknown	363
+GACGCTAT	Unknown		Unknown	357
+TAATCTAT	Unknown		Unknown	355
+CATTGCAT	Unknown		Unknown	355
+CAACAGAT	Unknown		Unknown	354
+GGAATCAT	Unknown		Unknown	353
+ACTTATAT	Unknown		Unknown	352
+CCCCCTCT	Unknown		Unknown	352
+GAATCGAT	Unknown		Unknown	351
+AAATCGAT	Unknown		Unknown	351
+CATTATCT	Unknown		Unknown	349
+CAGAGATC	Unknown		Unknown	348
+CAGGGCTG	Unknown		Unknown	348
+TAGAGGCA	N711		Unknown	347
+CAGGGCGA	Unknown		Unknown	347
+TCCTCCCT	Unknown		Unknown	346
+CCGACAAT	Unknown		Unknown	345
+GAAGGGGA	Unknown		Unknown	345
+CAGAGGAA	Unknown		Unknown	345
+GTAGAGCA	N712		Unknown	345
+TACGCTAT	Unknown		Unknown	344
+GCTAAGAT	Unknown		Unknown	343
+CGAGGTTG	N710		Unknown	342
+GTCTCTTC	Unknown		Unknown	342
+TAGAGATG	Unknown		Unknown	341
+TTATCGAT	Unknown		Unknown	341
+TACGCATC	Unknown		Unknown	341
+TCCCGTAT	Unknown		Unknown	341
+CATTCATC	Unknown		Unknown	339
+TGATCGAT	Unknown		Unknown	339
+CAGAGTAG	Unknown		Unknown	337
+CAACGAAT	Unknown		Unknown	337
+CACCGATC	Unknown		Unknown	337
+ATGCGCAT	N722		Unknown	336
+CATCTCGT	Unknown		Unknown	334
+CATGGATC	Unknown		Unknown	334
+CAGGGAAT	Unknown		Unknown	333
+CGAAAGAT	Unknown		Unknown	333
+ATTCCATC	Unknown		Unknown	333
+CACCCGAT	Unknown		Unknown	332
+GGACCTAT	Unknown		Unknown	332
+CATAGTAT	Unknown		Unknown	332
+ACATTCAT	Unknown		Unknown	331
+ACTCTTCC	Unknown		Unknown	331
+CTACCTAT	Unknown		Unknown	329
+GAAGCGGA	Unknown		Unknown	329
+CAAGCAAT	Unknown		Unknown	328
+CACCGCAT	Unknown		Unknown	328
+CAGACATG	Unknown		Unknown	327
+TCCCCGAA	D710		Unknown	326
+GGAATGCT	Unknown		Unknown	325
+AAGAGGTA	N711		Unknown	325
+TAATCATC	Unknown		Unknown	325
+CCTGCTAT	Unknown		Unknown	325
+CTCTCTCT	E502,S502,[N|S|E]502		Unknown	323
+GAGAGGGG	Unknown		Unknown	322
+CTAGCTAA	Unknown		Unknown	322
+ACTGTATC	Unknown		Unknown	322
+ATTTCTAT	Unknown		Unknown	321
+CAAGTAAT	Unknown		Unknown	320
+CAGAGGAG	Unknown		Unknown	319
+ATGAGCCA	Unknown		Unknown	318
+TAGGCCAT	Unknown		Unknown	318
+ACTCCTAT	Unknown		Unknown	318
+ATTAATAT	Unknown		Unknown	316
+CTAGAGGA	N712		Unknown	316
+ATGAGATC	Unknown		Unknown	316
+CCGGAATA	Unknown		Unknown	315
+CATAAGAT	Unknown		Unknown	315
+ATTCATCT	Unknown		Unknown	315
+CAGGCGGA	Unknown		Unknown	313
+TAGCCATG	N706		Unknown	312
+CAGCGAGT	Unknown		Unknown	311
+CAGCGGAT	Unknown		Unknown	310
+CTAGCTAC	Unknown		Unknown	309
+TAAATCAT	Unknown		Unknown	308
+GCCGCTAT	D502		Unknown	307
+TAACCGAT	Unknown		Unknown	307
+TACAGCCA	Unknown		Unknown	307
+CAGGCCGA	Unknown		Unknown	306
+CTTGCTAT	Unknown		Unknown	304
+GATAGGCT	Unknown		Unknown	304
+TCGGCCTG	Unknown		Unknown	302
+CAGATCTC	Unknown		Unknown	302
+GAGGGAGG	Unknown		Unknown	301
+CAGAGTAT	Unknown		Unknown	301
+TATGCATT	E506,S506,[N|S|E]506		Unknown	300
+AAGAGACG	Unknown		Unknown	299
+TAGTTCAT	Unknown		Unknown	297
+CCTCCTCT	Unknown		Unknown	297
+TAGGGAGG	Unknown		Unknown	296
+ATGATCAT	Unknown		Unknown	296
+AGAGGATG	E503,S503,[N|S|E]503		Unknown	294
+ACTATATC	Unknown		Unknown	292
+TACAATAT	Unknown		Unknown	292
+CATCCTAT	Unknown		Unknown	291
+CGAGTCTT	Unknown		Unknown	291
+GCCCGAAT	Unknown		Unknown	291
+CCAACTAT	Unknown		Unknown	290
+TACAGAAG	Unknown		Unknown	290
+CATAGAGT	S516		Unknown	289
+GACTACAT	Unknown		Unknown	288
+GACGCATC	Unknown		Unknown	288
+TTGAGCAT	Unknown		Unknown	287
+GTGAGCAT	Unknown		Unknown	285
+TCCCCAAT	Unknown		Unknown	285
+TAGGAAAG	Unknown		Unknown	284
+TCATACAT	Unknown		Unknown	284
+CAAAACAT	Unknown		Unknown	283
+GGCCTCAT	Unknown		Unknown	282
+CAGAGCAG	Unknown		Unknown	282
+TAGAGAAT	Unknown		Unknown	281
+TCTTCATC	Unknown		Unknown	280
+CATACTAT	Unknown		Unknown	280
+GGAATAAT	Unknown		Unknown	279
+ACGATATC	Unknown		Unknown	278
+TCCCGATA	Unknown		Unknown	277
+AAGAGAAG	Unknown		Unknown	277
+AGTCCTAT	Unknown		Unknown	276
+GTAGCGGT	Unknown		Unknown	276
+GACCACAT	Unknown		Unknown	276
+CCAAAGAT	Unknown		Unknown	275
+GTAGCTAT	Unknown		Unknown	274
+AGTGATAT	Unknown		Unknown	273
+CTTTTATC	Unknown		Unknown	273
+TATTCGAT	Unknown		Unknown	273
+GGACTAAT	Unknown		Unknown	273
+CACCGGAA	Unknown		Unknown	272
+GAAGAAGA	Unknown		Unknown	272
+CCATGGCA	Unknown		Unknown	271
+TCATTCCA	Unknown		Unknown	271
+TTAGCTAT	Unknown		Unknown	270
+TATACGAT	Unknown		Unknown	270
+CATCTAAT	S510		Unknown	269
+CATCGCAT	Unknown		Unknown	268
+GGAATACT	Unknown		Unknown	268
+AAGGCAAT	S520		Unknown	267
+ATTTCCTA	Unknown		Unknown	267
+ATGCCTAT	Unknown		Unknown	267
+TCTCCGCT	Unknown		Unknown	267
+CAGGAGAG	Unknown		Unknown	266
+GCTGCTAT	Unknown		Unknown	265
+TCGCGAAT	Unknown		Unknown	265
+ATCTCGTA	Unknown		Unknown	265
+GAGGAAGG	Unknown		Unknown	264
+GGACTTAT	Unknown		Unknown	264
+TATATTAT	Unknown		Unknown	263
+TTCCTATC	Unknown		Unknown	262
+CAAGCATG	Unknown		Unknown	261
+CAACTCCT	Unknown		Unknown	260
+CTTCTCCT	D505		Unknown	260
+AAGGCGCA	Unknown		Unknown	259
+CCACCAAT	Unknown		Unknown	259
+CCAAAAGA	Unknown		Unknown	258
+GCGACATC	Unknown		Unknown	257
+CAGAAAAT	Unknown		Unknown	256
+CATAGAAT	Unknown		Unknown	256
+ATCAGCAT	Unknown		Unknown	256
+AGGAGCAT	Unknown		Unknown	256
+CGGGGATG	Unknown		Unknown	256
+TACAGCAC	Unknown		Unknown	255
+CGAACAAT	Unknown		Unknown	254
+ACGGATAT	Unknown		Unknown	254
+CGATGCTG	N710		Unknown	253
+CAGTCTAT	Unknown		Unknown	252
+CTAGCTCT	Unknown		Unknown	251
+CTTGGCAT	Unknown		Unknown	251
+TAGAGGAT	Unknown		Unknown	251
+CAAAGCAT	Unknown		Unknown	251
+CTCGCTAT	E502,S502,[N|S|E]502		Unknown	250
+CCGGCGCT	Unknown		Unknown	250
+GCGCTATC	Unknown		Unknown	249
+AAGAGAAT	Unknown		Unknown	248
+CACAGAGG	N708		Unknown	247
+TATTATAT	Unknown		Unknown	247
+TCCCAAAT	Unknown		Unknown	247
+CTCTCTTC	N707		Unknown	246
+CATTCCTA	Unknown		Unknown	246
+GGAGTCAT	Unknown		Unknown	245
+GGAGACAT	Unknown		Unknown	245
+CAATTAAT	Unknown		Unknown	244
+CTATAATA	Unknown		Unknown	243
+ATTCCTAC	Unknown		Unknown	243
+GATCTCGT	Unknown		Unknown	242
+GAGAGGAT	Unknown		Unknown	242
+TAGGCGTT	Unknown		Unknown	241
+CATTCGAT	Unknown		Unknown	241
+CACCGTAT	Unknown		Unknown	241
+AAGGCGAA	Unknown		Unknown	241
+CAACAAAT	Unknown		Unknown	241
+CTAGGTAT	Unknown		Unknown	240
+CAAGGAGG	Unknown		Unknown	240
+TAGACGAT	Unknown		Unknown	240
+CGAGGATC	Unknown		Unknown	239
+GAGAGGCT	Unknown		Unknown	238
+GAAGCGAT	Unknown		Unknown	238
+TGACTCAT	Unknown		Unknown	238
+TAGGCTTG	N706		Unknown	237
+GGGATCCT	Unknown		Unknown	237
+TACAGCGT	Unknown		Unknown	237
+CGGGAGAT	Unknown		Unknown	237
+TCCCTAAT	Unknown		Unknown	236
+CAGAGAGC	N708		Unknown	236
+TACTCGAT	Unknown		Unknown	236
+CAGGGCAG	Unknown		Unknown	236
+CTCACTAT	E502,S502,[N|S|E]502		Unknown	236
+GCCACGAT	Unknown		Unknown	236
+GGAGACCT	Unknown		Unknown	236
+GACCGGAT	Unknown		Unknown	235
+TAGACAGG	Unknown		Unknown	235
+CCCTCTAC	N707		Unknown	235
+TACTTCAT	Unknown		Unknown	234
+CAAAGAAG	Unknown		Unknown	234
+GTAGGAAT	Unknown		Unknown	234
+CAGAGGCG	Unknown		Unknown	234
+CACTCTAT	E502,S502,[N|S|E]502		Unknown	233
+TATTCATC	Unknown		Unknown	233
+TCATTTCA	Unknown		Unknown	233
+GAAGAAAT	Unknown		Unknown	233
+AACAGGCA	N711		Unknown	233
+CGAGGAGG	Unknown		Unknown	232
+GGATTCAT	Unknown		Unknown	232
+GGAGAGGG	Unknown		Unknown	231
+CAGTTTAT	Unknown		Unknown	231
+ATATATAA	Unknown		Unknown	231
+CCCGAATC	Unknown		Unknown	230
+CAGGAGAA	Unknown		Unknown	229
+TCCTATCT	Unknown		Unknown	228
+CAGGCGCA	Unknown		Unknown	228
+GGAACGCT	Unknown		Unknown	227
+TAACTCAT	Unknown		Unknown	227
+GACGTCAT	Unknown		Unknown	227
+TATCTCGT	Unknown		Unknown	226
+TCATGCAT	Unknown		Unknown	226
+CAGAAAAG	Unknown		Unknown	225
+ATTCTCCT	Unknown		Unknown	225
+ATAATATC	Unknown		Unknown	225
+CAGAGACA	Unknown		Unknown	225
+GCGACGCG	Unknown		Unknown	225
+CAGAGGCT	Unknown		Unknown	224
+ACGACATC	Unknown		Unknown	224
+CAGGATAT	Unknown		Unknown	224
+AAGACGCT	Unknown		Unknown	224
+ACTTCTCT	Unknown		Unknown	224
+TACACGAT	Unknown		Unknown	223
+CACGCCAT	Unknown		Unknown	223
+CAAGGCAG	Unknown		Unknown	222
+GAGGCATT	Unknown		Unknown	220
+CTCATTCA	Unknown		Unknown	220
+TAATCGGT	Unknown		Unknown	220
+TCAGCATC	Unknown		Unknown	219
+CAGTCTAC	Unknown		Unknown	219
+ACTGTTAT	Unknown		Unknown	219
+ACCCGAAT	Unknown		Unknown	219
+TAAAGGAT	Unknown		Unknown	219
+GGAGGCTT	Unknown		Unknown	218
+CTTACTAT	Unknown		Unknown	217
+CACGGGAT	Unknown		Unknown	217
+CTACAGCA	Unknown		Unknown	216
+GAAGCATG	Unknown		Unknown	215
+TAGGCCTT	Unknown		Unknown	215
+GAACTACT	Unknown		Unknown	214
+CAGGCATT	Unknown		Unknown	214
+CAACTAGT	Unknown		Unknown	214
+GACGACCA	Unknown		Unknown	213
+CCTACAAT	Unknown		Unknown	213
+CTGATATC	Unknown		Unknown	213
+TATTTCAT	Unknown		Unknown	212
+ATCCCTAT	Unknown		Unknown	212
+TAAGCATT	Unknown		Unknown	211
+CAGCGAAT	Unknown		Unknown	210
+TTCCGAAT	Unknown		Unknown	210
+ACCGATAT	Unknown		Unknown	210
+GACGACGT	Unknown		Unknown	210
+CAGACGGT	Unknown		Unknown	209
+CAACCGAT	Unknown		Unknown	209
+TCCATTCA	Unknown		Unknown	208
+GTAGCGAA	Unknown		Unknown	208
+TATAATGT	Unknown		Unknown	208
+CTCCCTAT	E502,S502,[N|S|E]502		Unknown	207
+GCAAAGCT	Unknown		Unknown	207
+GACGAAAG	Unknown		Unknown	206
+CAAGCCAT	Unknown		Unknown	206
+TAATCGAC	Unknown		Unknown	205
+GGAGACGA	Unknown		Unknown	205
+CAGATAGT	Unknown		Unknown	205
+CCTAGCTA	Unknown		Unknown	205
+CATGGCCA	Unknown		Unknown	205
+CATTTTTT	Unknown		Unknown	204
+CCAGCGAT	Unknown		Unknown	204
+CCTACTAT	Unknown		Unknown	204
+GCAACACT	Unknown		Unknown	203
+ACTCATAT	Unknown		Unknown	202
+GTAGATAA	Unknown		Unknown	202
+GATGATAT	Unknown		Unknown	202
+CATGAGAT	Unknown		Unknown	202
+GACACATC	Unknown		Unknown	202
+TGAGCATC	Unknown		Unknown	201
+CCGACGCT	Unknown		Unknown	201
+GTACAGGA	N712		Unknown	200
+CGAGGCCT	Unknown		Unknown	200
+GTAGTGGA	N712		Unknown	200
+CAAGTCAT	Unknown		Unknown	200
+CTCTCTAG	E502,N707,S502,[N|S|E]502		Unknown	200
+TAGGTGAT	Unknown		Unknown	200
+GAGAGAGT	Unknown		Unknown	199
+GCAGCGCT	Unknown		Unknown	198
+AGGCGATA	D501		Unknown	198
+CCCCTTCT	Unknown		Unknown	198
+CATCGGAT	Unknown		Unknown	197
+CTAGTTAT	Unknown		Unknown	197
+TAATCGTT	Unknown		Unknown	196
+TATACGCT	Unknown		Unknown	196
+GGTCCTAT	Unknown		Unknown	195
+TAAACAAT	Unknown		Unknown	195
+GCTCCTAT	Unknown		Unknown	195
+GGAAGCCT	Unknown		Unknown	195
+CGACGCTG	N710		Unknown	195
+TAAGTCAT	Unknown		Unknown	194
+TATAAATA	Unknown		Unknown	194
+CATTTTAC	Unknown		Unknown	193
+CTAATCGA	A505		Unknown	193
+GTTACGAT	Unknown		Unknown	193
+CACTGATA	Unknown		Unknown	193
+TCCTGAAT	Unknown		Unknown	193
+CAACTTAT	Unknown		Unknown	192
+CACAGAAT	Unknown		Unknown	192
+GGACTATC	Unknown		Unknown	192
+CATGAGCA	Unknown		Unknown	192
+CGGATCTC	Unknown		Unknown	192
+CTAGTATC	Unknown		Unknown	192
+ATAGCATC	Unknown		Unknown	192
+GCCTCCCT	Unknown		Unknown	192
+GTACTCAT	Unknown		Unknown	191
+ACCTGATA	Unknown		Unknown	191
+GAGGAATG	Unknown		Unknown	191
+TAAGGCAT	Unknown		Unknown	191
+CACCAGAT	Unknown		Unknown	190
+TAATATCT	Unknown		Unknown	189
+CCAACGCT	Unknown		Unknown	188
+CACTTTAT	Unknown		Unknown	187
+ATCTCTAA	Unknown		Unknown	187
+GGGCTACT	Unknown		Unknown	187
+TCTACGAT	Unknown		Unknown	186
+GTCTCTAT	D502,E502,S502,[N|S|E]502		Unknown	186
+ATGAGCAC	Unknown		Unknown	186
+TCATTCAC	Unknown		Unknown	185
+CGAGCTAT	Unknown		Unknown	185
+TATGCTAT	Unknown		Unknown	185
+TACTGCAT	Unknown		Unknown	185
+CGAGGAAG	Unknown		Unknown	185
diff --git a/test/input/TestIlluminaBarcodeHelper/single_index/expected.txt b/test/input/TestIlluminaBarcodeHelper/single_index/expected.txt
new file mode 100644
index 000000000..57f3b5e2c
--- /dev/null
+++ b/test/input/TestIlluminaBarcodeHelper/single_index/expected.txt
@@ -0,0 +1,5 @@
+sample_name	expected_barcode_1	expected_barcode_1_name	guessed_barcode_1	guessed_barcode_1_name	match_type
+EBOV-VP24UTR_G10218A-C10220U.DMSO	CAAAAGAT				alternative_indices_uncertain
+EBOV-VP24UTR_G10218A-C10242U.DMSO	CACCGGAT				alternative_indices_uncertain
+EBOV-VP24UTR_G10218A-C10242U.NMIA	CTAGCTAT				alternative_indices_uncertain
+EBOV-VP24UTR_G10218A-C10242U.DENAT	TCCCGAAT				alternative_indices_uncertain
diff --git a/test/input/TestIlluminaBarcodeHelper/single_index/metrics.txt b/test/input/TestIlluminaBarcodeHelper/single_index/metrics.txt
new file mode 100644
index 000000000..2b764f1e7
--- /dev/null
+++ b/test/input/TestIlluminaBarcodeHelper/single_index/metrics.txt
@@ -0,0 +1,36 @@
+## htsjdk.samtools.metrics.StringHeader
+# ExtractIlluminaBarcodes BASECALLS_DIR=tmp/00_flowcells/171110_M04004_0212_000000000-BDTT8/Data/Intensities/BaseCalls OUTPUT_DIR=/tmp/8574290.1.broad/tmp-illumina-illumina_demux-8574290-0-5ctjr6gn/extracted_barcodes-qt71mddp LANE=1 READ_STRUCTURE=150T8B150T BARCODE_FILE=/tmp/8574290.1.broad/tmp-illumina-illumina_demux-8574290-0-5ctjr6gn/barcodeData.BDTT8.1x4fc_7rw.txt METRICS_FILE=/idi/sabeti-scratch/alin/software/viral-ngs-etc/projects/190206-LCMV-C2c2/reports/barcodes/barcodes-metrics-BDTT8.1.txt MAX_MISMATCHES=0 MINIMUM_BASE_QUALITY=20 NUM_PROCESSORS=0    MIN_MISMATCH_DELTA=1 MAX_NO_CALLS=2 MINIMUM_QUALITY=2 COMPRESS_OUTPUTS=false VERBOSITY=INFO QUIET=false VALIDATION_STRINGENCY=STRICT COMPRESSION_LEVEL=5 MAX_RECORDS_IN_RAM=500000 CREATE_INDEX=false CREATE_MD5_FILE=false GA4GH_CLIENT_SECRETS=client_secrets.json USE_JDK_DEFLATER=false USE_JDK_INFLATER=false
+## htsjdk.samtools.metrics.StringHeader
+# Started on: Sat Feb 16 00:09:44 EST 2019
+
+## METRICS CLASS	picard.illumina.ExtractIlluminaBarcodes$BarcodeMetric
+BARCODE	BARCODE_WITHOUT_DELIMITER	BARCODE_NAME	LIBRARY_NAME	READS	PF_READS	PERFECT_MATCHES	PF_PERFECT_MATCHES	ONE_MISMATCH_MATCHES	PF_ONE_MISMATCH_MATCHES	PCT_MATCHES	RATIO_THIS_BARCODE_TO_BEST_BARCODE_PCT	PF_PCT_MATCHES	PF_RATIO_THIS_BARCODE_TO_BEST_BARCODE_PCT	PF_NORMALIZED_MATCHES
+GGACTCCT	GGACTCCT	pTOPO-LCMV-Arm-emGFP-S-NoEnds_P3-IC-170825-cloned_171103	pTOPO-LCMV-Arm-emGFP-S-NoEnds_P3-IC-170825-cloned_171103.l1	1282976	1239382	1282976	1239382	0	0	0.060095	0.655219	0.065356	0.652624	1.904065
+TAGGCATG	TAGGCATG	pcDNA34-MGC-ADAR2-V5	pcDNA34-MGC-ADAR2-V5.l1	1192766	1159442	1192766	1159442	0	0	0.055869	0.609149	0.06114	0.61053	1.781253
+CTCTCTAC	CTCTCTAC	pT7CFE-MGC-ADAR2	pT7CFE-MGC-ADAR2.l1	1051975	1009354	1051975	1009354	0	0	0.049275	0.537247	0.053226	0.531497	1.550672
+CAGAGAGG	CAGAGAGG	p20_3E5E_fLuc2	p20_3E5E_fLuc2.l1	1958086	1899076	1958086	1899076	0	0	0.091717	1	0.100143	1	2.917554
+GCTACGCT	GCTACGCT	p20_3E5E_rLuc-fLuc2_6	p20_3E5E_rLuc-fLuc2_6.l1	1184157	1158385	1184157	1158385	0	0	0.055466	0.604752	0.061084	0.609973	1.779629
+CGAGGCTG	CGAGGCTG	p20_3E5E_rLuc-fLuc2_5	p20_3E5E_rLuc-fLuc2_5.l1	941504	917972	941504	917972	0	0	0.0441	0.480829	0.048407	0.483378	1.410282
+AAGAGGCA	AAGAGGCA	p20_3E5E_rLuc-fLuc2_7	p20_3E5E_rLuc-fLuc2_7.l1	765762	744083	765762	744083	0	0	0.035868	0.391077	0.039237	0.391813	1.143136
+GTAGAGGA	GTAGAGGA	p20_3E5E_rLuc-fLuc2_8	p20_3E5E_rLuc-fLuc2_8.l1	940017	917850	940017	917850	0	0	0.044031	0.480069	0.0484	0.483314	1.410095
+ACTGATAT	ACTGATAT	EBOV-VP24UTR_G10218.DMSO	EBOV-VP24UTR_G10218.DMSO.l1	538711	517696	538711	517696	0	0	0.025233	0.275121	0.027299	0.272604	0.795337
+ATGAGCAT	ATGAGCAT	EBOV-VP24UTR_G10218A.DMSO	EBOV-VP24UTR_G10218A.DMSO.l1	382966	366987	382966	366987	0	0	0.017938	0.195582	0.019352	0.193245	0.563803
+ATTCCTAT	ATTCCTAT	EBOV-VP24UTR_G10218A-A10232G.DMSO	EBOV-VP24UTR_G10218A-A10232G.DMSO.l1	490275	472589	490275	472589	0	0	0.022965	0.250385	0.024921	0.248852	0.726039
+CAAAAGAT	CAAAAGAT	EBOV-VP24UTR_G10218A-C10220U.DMSO	EBOV-VP24UTR_G10218A-C10220U.DMSO.l1	264750	255281	264750	255281	0	0	0.012401	0.135209	0.013462	0.134424	0.392189
+CAACTAAT	CAACTAAT	EBOV-VP24UTR_G10218A-A10240G.DMSO	EBOV-VP24UTR_G10218A-A10240G.DMSO.l1	353697	337630	353697	337630	0	0	0.016567	0.180634	0.017804	0.177786	0.518702
+CACCGGAT	CACCGGAT	EBOV-VP24UTR_G10218A-C10242U.DMSO	EBOV-VP24UTR_G10218A-C10242U.DMSO.l1	187070	179344	187070	179344	0	0	0.008762	0.095537	0.009457	0.094438	0.275526
+CAGGCGAT	CAGGCGAT	EBOV-VP24UTR_G10218.NMIA	EBOV-VP24UTR_G10218.NMIA.l1	1179793	1125673	1179793	1125673	0	0	0.055262	0.602524	0.059359	0.592748	1.729373
+CATGGCAT	CATGGCAT	EBOV-VP24UTR_G10218A.NMIA	EBOV-VP24UTR_G10218A.NMIA.l1	386930	371948	386930	371948	0	0	0.018124	0.197606	0.019614	0.195857	0.571424
+CATTTTAT	CATTTTAT	EBOV-VP24UTR_G10218A-A10232G.NMIA	EBOV-VP24UTR_G10218A-A10232G.NMIA.l1	561855	537740	561855	537740	0	0	0.026317	0.286941	0.028356	0.283159	0.826131
+CCAACAAT	CCAACAAT	EBOV-VP24UTR_G10218A-C10220U.NMIA	EBOV-VP24UTR_G10218A-C10220U.NMIA.l1	395259	379142	395259	379142	0	0	0.018514	0.20186	0.019993	0.199646	0.582477
+CGGAATAT	CGGAATAT	EBOV-VP24UTR_G10218A-A10240G.NMIA	EBOV-VP24UTR_G10218A-A10240G.NMIA.l1	487124	470904	487124	470904	0	0	0.022817	0.248776	0.024832	0.247965	0.723451
+CTAGCTAT	CTAGCTAT	EBOV-VP24UTR_G10218A-C10242U.NMIA	EBOV-VP24UTR_G10218A-C10242U.NMIA.l1	268529	257633	268529	257633	0	0	0.012578	0.137139	0.013586	0.135662	0.395802
+GACGACAT	GACGACAT	EBOV-VP24UTR_G10218.DENAT	EBOV-VP24UTR_G10218.DENAT.l1	471442	454685	471442	454685	0	0	0.022082	0.240767	0.023977	0.239424	0.698533
+TAATCGAT	TAATCGAT	EBOV-VP24UTR_G10218A.DENAT	EBOV-VP24UTR_G10218A.DENAT.l1	349709	334453	349709	334453	0	0	0.01638	0.178597	0.017637	0.176114	0.513821
+TACAGCAT	TACAGCAT	EBOV-VP24UTR_G10218A-A10232G.DENAT	EBOV-VP24UTR_G10218A-A10232G.DENAT.l1	588492	559195	588492	559195	0	0	0.027565	0.300545	0.029488	0.294456	0.859092
+TATAATAT	TATAATAT	EBOV-VP24UTR_G10218A-C10220U.DENAT	EBOV-VP24UTR_G10218A-C10220U.DENAT.l1	482453	460095	482453	460095	0	0	0.022598	0.24639	0.024262	0.242273	0.706845
+TCATTCAT	TCATTCAT	EBOV-VP24UTR_G10218A-A10240G.DENAT	EBOV-VP24UTR_G10218A-A10240G.DENAT.l1	541051	519095	541051	519095	0	0	0.025343	0.276316	0.027373	0.273341	0.797487
+TCCCGAAT	TCCCGAAT	EBOV-VP24UTR_G10218A-C10242U.DENAT	EBOV-VP24UTR_G10218A-C10242U.DENAT.l1	287559	278125	287559	278125	0	0	0.013469	0.146857	0.014666	0.146453	0.427284
+NNNNNNNN	NNNNNNNN			3814272	2039896	0	0	0	0	0.178661	1.947959	0.107569	1.074152	0
+
+
diff --git a/test/unit/test_illumina.py b/test/unit/test_illumina.py
index 30e08753c..d4df77b2e 100644
--- a/test/unit/test_illumina.py
+++ b/test/unit/test_illumina.py
@@ -264,6 +264,20 @@ def test_ambiguous(self):
 
         self.assertEqualContents(out_report, expected)
 
+    def test_single_index_run(self):
+        dir_prefix = "single_index"
+        in_dir = util.file.get_test_input_path(self)
+        in_barcodes = os.path.join(in_dir,dir_prefix,"barcodes.txt")
+        in_metrics = os.path.join(in_dir,dir_prefix,"metrics.txt")
+        out_report = util.file.mkstempfname('.txt')
+        expected = os.path.join(in_dir,dir_prefix,"expected.txt")
+
+        args = [in_barcodes, in_metrics, out_report]
+        args = illumina.parser_guess_barcodes(argparse.ArgumentParser()).parse_args(args)
+        args.func_main(args)
+
+        self.assertEqualContents(out_report, expected)
+
     def test_few_assigned(self):
         dir_prefix = "few_assigned"
         in_dir = util.file.get_test_input_path(self)
diff --git a/util/illumina_indices.py b/util/illumina_indices.py
index cdb9e9d3b..554147c7f 100755
--- a/util/illumina_indices.py
+++ b/util/illumina_indices.py
@@ -576,7 +576,7 @@ def __init__(self, barcode_counts, picard_metrics, sample_name, rows_limit=1000)
         for row in util.file.read_tabfile_dict(picard_metrics, skip_prefix="#"):
             barcodes = tuple(row["BARCODE"].split("-"))
             if "BARCODE_NAME" in row:
-                self.sample_to_barcodes[row["BARCODE_NAME"]]       = barcodes
+                self.sample_to_barcodes[row["BARCODE_NAME"]] = barcodes
                 self.samples.append(row["BARCODE_NAME"])
                 self.sample_to_read_counts[row["BARCODE_NAME"]] = int(row["READS"])
             elif all(re.match(r'^N+$',barcode) for barcode in barcodes):
@@ -587,9 +587,11 @@ def __init__(self, barcode_counts, picard_metrics, sample_name, rows_limit=1000)
         #Barcode1   Likely_Index_Names1 Barcode2    Likely_Index_Names2 Count
         #CTCTCTAC   N707    AAGGAGTA    S507,[N|S|E]507 40324834
         for row in util.file.read_tabfile_dict(barcode_counts, rowcount_limit=rows_limit):
-            self.barcodes_seen[(row["Barcode1"],row["Barcode2"])] = int(row["Count"])
+            if (row["Barcode1"],row.get("Barcode2",None)) not in self.barcodes_seen:
+                self.barcodes_seen[(row["Barcode1"],row.get("Barcode2",None))] = int(row["Count"])
             self.barcode_name_map[row["Barcode1"]] = row["Likely_Index_Names1"]
-            self.barcode_name_map[row["Barcode2"]] = row["Likely_Index_Names2"]
+            if "Barcode2" in row and row["Barcode2"]:
+                self.barcode_name_map[row["Barcode2"]] = row["Likely_Index_Names2"]
 
     def outlier_barcodes(self, outlier_threshold=0.675, expected_assigned_fraction=0.7, number_of_negative_controls=1):
         """
@@ -711,30 +713,35 @@ def guess_barcodes_for_sample(self, sample_name):
             else:
                 del barcodes_seen_novel[barcode_pair]
 
+        is_dual_index = len(self.sample_to_barcodes[sample_name]) > 1
+
         out_dict["expected_barcode_1"]           = self.sample_to_barcodes[sample_name][0]
         out_dict["expected_barcode_1_name"]      = ",".join(self.index_reference.guess_index(self.sample_to_barcodes[sample_name][0]))
-        out_dict["expected_barcode_2"]           = self.sample_to_barcodes[sample_name][1]
-        out_dict["expected_barcode_2_name"]      = ",".join(self.index_reference.guess_index(self.sample_to_barcodes[sample_name][1]))
-        out_dict["expected_barcodes_read_count"] = self.sample_to_read_counts[sample_name]
+        if is_dual_index:
+            out_dict["expected_barcode_2"]           = self.sample_to_barcodes[sample_name][1]
+            out_dict["expected_barcode_2_name"]      = ",".join(self.index_reference.guess_index(self.sample_to_barcodes[sample_name][1]))
+            out_dict["expected_barcodes_read_count"] = self.sample_to_read_counts[sample_name]
 
         claimed_barcodes = self.sample_to_barcodes[sample_name]
 
         found_partial_match = False
         putative_match = None
-        # barcodes_seen_novel is sorted by read count, desc
-        for (barcode_pair,count) in barcodes_seen_novel.items():
-            if barcode_pair[0]==claimed_barcodes[0] or barcode_pair[1]==claimed_barcodes[1]:
-                found_partial_match=True
-                putative_match = barcode_pair
-                out_dict["match_type"] = "one_barcode_match"
-                break
 
-        # find index of match to help determine if it is a reasonable guess
-        idx_of_match = -1
-        for (idx,(barcode_pair,count)) in enumerate(self.barcodes_seen.items()):
-            if barcode_pair==putative_match:
-                idx_of_match=idx
-                break
+        if is_dual_index:
+        # barcodes_seen_novel is sorted by read count, desc
+            for (barcode_pair,count) in barcodes_seen_novel.items():
+                if barcode_pair[0]==claimed_barcodes[0] or barcode_pair[1]==claimed_barcodes[1]:
+                    found_partial_match=True
+                    putative_match = barcode_pair
+                    out_dict["match_type"] = "one_barcode_match"
+                    break
+
+            # find index of match to help determine if it is a reasonable guess
+            idx_of_match = -1
+            for (idx,(barcode_pair,count)) in enumerate(self.barcodes_seen.items()):
+                if barcode_pair==putative_match:
+                    idx_of_match=idx
+                    break
 
         # if the one-barcode match is too far down the list of barcode pairs seen
         # (farther down than 1.5x the number of samples)
@@ -747,9 +754,10 @@ def guess_barcodes_for_sample(self, sample_name):
         
         out_dict["guessed_barcode_1"]           = putative_match[0]
         out_dict["guessed_barcode_1_name"]      = self.barcode_name_map[putative_match[0]]
-        out_dict["guessed_barcode_2"]           = putative_match[1]
-        out_dict["guessed_barcode_2_name"]      = self.barcode_name_map[putative_match[1]]
-        out_dict["guessed_barcodes_read_count"] = self.barcodes_seen[(putative_match[0],putative_match[1])]
+        if is_dual_index:
+            out_dict["guessed_barcode_2"]           = putative_match[1]
+            out_dict["guessed_barcode_2_name"]      = self.barcode_name_map[putative_match[1]]
+            out_dict["guessed_barcodes_read_count"] = self.barcodes_seen[(putative_match[0],putative_match[1])]
 
         return out_dict
 
@@ -781,7 +789,7 @@ def find_uncertain_barcodes(self, sample_names=None, outlier_threshold=0.675, ex
         consolidated_guesses = defaultdict(list)
 
         for row in guessed_barcodes:
-            consolidated_guesses[(row["guessed_barcode_1"],row["guessed_barcode_2"])].append(row)
+            consolidated_guesses[(row["guessed_barcode_1"],row.get("guessed_barcode_2",None))].append(row)
 
         final_guesses = []
 
@@ -790,7 +798,8 @@ def clear_guessed_fields(sample,match_reason,fields_to_clear=None):
                                                     "guessed_barcode_1_name", "guessed_barcode_2_name",
                                                     "guessed_barcodes_read_count"]
             for field in fields_to_clear:
-                sample[field] = None
+                if field in sample:
+                    sample[field] = None
             sample["match_type"] = match_reason
             return sample
 
@@ -798,7 +807,7 @@ def clear_guessed_fields(sample,match_reason,fields_to_clear=None):
             if len(samples) > 1:
                 log.warning("Ambiguous! Multiple samples corresponding to guessed barcodes %s:", barcode_pair)
                 for sample in samples:
-                    log.warning("\t%s expected (%s,%s); -> Guessed (%s,%s); match type: %s", sample["sample_name"], sample["expected_barcode_1"],sample["expected_barcode_2"],sample["guessed_barcode_1"],sample["guessed_barcode_2"],sample["match_type"])
+                    log.warning("\t%s expected (%s,%s); -> Guessed (%s,%s); match type: %s", sample["sample_name"], sample["expected_barcode_1"],sample.get("expected_barcode_2",""),sample["guessed_barcode_1"],sample.get("guessed_barcode_2",""),sample["match_type"])
     
                     final_guesses.append(clear_guessed_fields(sample, "alternative_indices_uncertain"))
             else:
@@ -807,11 +816,10 @@ def clear_guessed_fields(sample,match_reason,fields_to_clear=None):
                         final_guesses.append(clear_guessed_fields(sample, "alternatives_have_lower_read_counts"))
                     else:
                         final_guesses.append(sample)
-
         return final_guesses
 
     def write_guessed_barcodes(self, out_tsv, guessed_barcodes):
-        output_header = ["sample_name",
+        possible_header_cols = ["sample_name",
                             "expected_barcode_1","expected_barcode_2",
                             "expected_barcode_1_name","expected_barcode_2_name",
                             "expected_barcodes_read_count",
@@ -820,10 +828,15 @@ def write_guessed_barcodes(self, out_tsv, guessed_barcodes):
                             "guessed_barcodes_read_count",
                             "match_type"
                         ]
+        output_header_columns = []
+        for header_key in possible_header_cols:
+            for row in guessed_barcodes:
+                if header_key in row.keys() and header_key not in output_header_columns:
+                    output_header_columns.append(header_key)
 
         with open(out_tsv, 'w') as tsvfile:
             csv.register_dialect('dict_tsv', quoting=csv.QUOTE_MINIMAL, delimiter="\t")
-            writer = csv.DictWriter(tsvfile, fieldnames=output_header, dialect="dict_tsv")
+            writer = csv.DictWriter(tsvfile, fieldnames=output_header_columns, dialect="dict_tsv")
 
             writer.writeheader()
             writer.writerows(guessed_barcodes)