same reads with different CIGAR

[vincent lee]

Hi all,
I am using bwa(Version: 0.6.2-r126, default parameter) for reads alignment, and convert the result into bam format.
But I find some reads with same sequence have different CIGAR value. Would someone can explain this?

ILLUMINA-8C6145:1209:MP-PE7:7:30:4170:19068     163     chr12_pos       566     29      81M19S  =       647     203     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCA    gchhgghhhhdhghhhghfghfgffghfcfggfgdfchhcfdggffedfdddbbbd`fffadddadaaad][\aaaa```aeccccc[aaaccca_a]aa    XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A54
ILLUMINA-8C6145:1209:MP-PE7:7:85:17609:15405    99      chr12_pos       566     29      52M70S  =       618     152     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCATTTATCTTGAATCTAATTATAC      hhghhhhfhhfhhhhgfhefhaggeghgghghhhhhhhhhehhhfhefhfggeeghggfgcggfgddagdbeaddfdeegfhgbfffcfbeccfdafcffd`ccdccccccdccccdc_d^_      XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A25
ILLUMINA-8C6145:1209:MP-PE7:7:103:19002:21043   163     chr12_pos       566     29      30M70S  =       596     152     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCA    hghhhdhh]hgfehghggg[acacaeggggg_acbeedeghgghfhbdffedbbbgbddcYaa_aaaaaa_b]bba```b_ca]__a]_baaaaYcaadc    XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A3
ILLUMINA-8C6145:1209:MP-PE7:7:106:13284:18011   163     chr12_pos       566     29      56M44S  =       622     178     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCA    hhhhfhhhhghhhhghhhhhhhhgghhhghhhhhghhhhhghhhhgcggggggghhefggahggcedgdgdhadWddcbbgfgdfdhdfagfcfcdadab    XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A29
ILLUMINA-8C6145:1019:MP-PE6:7:33:7888:15225     163     chr12_pos       566     29      98M2S   =       664     220     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCA    hhhhhghhhhgghhhhhghhfgchhhhhhhhhfeghedLhfffccggghhggggdcgeegahfgfhgfffaead_deebbeffagcdabae]a_d_bbba    XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A71
ILLUMINA-8C6145:1019:MP-PE6:7:50:3845:1687      99      chr12_pos       566     29      42M80S  =       608     142     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCATTTATCTTGAATCTAATTATAC      hhghhghhhhhhhhhgfghhhhghhfhggghhhhhfhfhhhghhhchghhhhhhghhhhhdehhhhhhghhhfghhgggehghghhgdffghfgfhhfhffgfgfgfff_cddddcfdfffc      XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A15
ILLUMINA-8C6145:1019:MP-PE6:7:116:1929:9674     163     chr12_pos       566     29      39M61S  =       605     161     ATGCCTTATCGCTTAGCGTTGTTGTCGAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCA    hghgghhdhhghhhhhhhhhhhhhgeheeghcghhdegfggghchcdghgaggghhbdghahhdcddaddT[W[I^\]\\dggegegbfebbabYbdba_    XT:A:M  NM:i:3  SM:i:29 AM:i:29 XM:i:3  XO:i:0  XG:i:0  MD:Z:9T11C4A12

The reference sequence(chr12_pos 566-666):

ATGCCTTATTGCTTAGCGTTGCTGTCAAAAAATATGATCCAAAATCGAGAAAAGAACAAGTATCCATCCAAGTTTTCTTACTAATTAAAATGAACTTCCAT

Thanks
Lee

[dpryan]

My only guess (I don't use bwa much) would be that it has something to do with the mapping of the mate read. The XT tag is saying that there was a mate rescue, so perhaps it's soft-clipping where the mates overlap? If no one else posts with better ideas, then try posting a couple of the mates so we can see if there's an interaction between the mates position and the soft-clipping.

[vincent lee]

My only guess (I don't use bwa much) would be that it has something to do with the mapping of the mate read. The XT tag is saying that there was a mate rescue, so perhaps it's soft-clipping where the mates overlap? If no one else posts with better ideas, then try posting a couple of the mates so we can see if there's an interaction between the mates position and the soft-clipping.

Thank you, dpryan.
May be you get the correct answer. I have checked the source code of samtools. If using M instead of S, the overlap part of read will be counted twice in mpileup. I am not sure whether it may casuse some problems...

Lee