Edit distance in BWA

[Rachelly]

Hi all,
I'm using BWA for alignment and want to get only alignments with up to 1 mismatch/ indel per read.
So I've set the -n flag to 1, but I still get reads with a bigger edit distance..

SBS123:68:C00PFABXX:3:1104:7837:18918   0       MED4_genome     371700  37      2M1D48M *       0   0GAAAAAAAAAATGTAAAATATGGAACTGAATTTTTCGGAATTAATAGAGC      CCCFFFFFHHHHGHIJJIGGGHGIHIGIJIIJJJHIJIIIJIIGIIIIHF   XT:A:U  NM:i:2  X0:i:1  X1:i:0  XM:i:0  XO:i:1  XG:i:1  MD:Z:0A1^G48
SBS123:68:C00PFABXX:3:1106:12571:108775 20      MED4_genome     1657990 25      50M     *       0   0TTGATGGTTAACAGAAATAAGAAGGTGGAAAAAAAAGCATAAATGTTGAT      FB8FDGHEHIFHEDEFB*>HDDIGIGGIIIGHFFDHGHHHFFD?FFF@@@   XT:A:U  NM:i:28 X0:i:1  X1:i:0  XM:i:1  XO:i:0  XG:i:0  MD:Z:0A0A0A1A0A0A0A0A2A1A3A2A2A0A0A0A0A0G1C5A0A1A3A0A0A0A0A1A0
SBS123:68:C00PFABXX:3:1106:19540:193836 20      MED4_genome     1657990 25      50M     *       0   0TTGATGGTTAACAGAAATAAGAAGGTGGAAAAAAAAGCATAAATGTTGAT      EFFC83BFFB@F?<DB<9F?*F?)?C:9:EFFFFCBF<?<FADDADD:1@   XT:A:U  NM:i:28 X0:i:1  X1:i:0  XM:i:1  XO:i:0  XG:i:0  MD:Z:0A0A0A1A0A0A0A0A2A1A3A2A2A0A0A0A0A0G1C5A0A1A3A0A0A0A0A1A0
SBS123:68:C00PFABXX:3:1107:10937:66747  0       MED4_genome     1494995 37      1M1D49M *       0   0CCCCTTTTTTTTTAATGAATCTTCTAAAGCATCACTTAAAGTTTGCATTG      @@@DABD>FDFDFBB?D>B*19?BDHCBH4B<?*9?BAHHIBH@FHDFGH   XT:A:U  NM:i:2  X0:i:1  X1:i:0  XM:i:0  XO:i:1  XG:i:1  MD:Z:1^T3C45
SBS123:68:C00PFABXX:3:1108:2389:8959    0       MED4_genome     531215  37      3M1D47M *       0   0

I saw an old thread about this issue, but no conclusions are written there:

http://seqanswers.com/forums/showthread.php?t=3044

What might cause this behavior of BWA?

Thanks,
Rachelly.

[nupurgupta]

BWA edit distance

Hi,
Did you find a solution for this at all? I am getting the same problem.

[Rachelly]

Hi,
Unfortunately I didn't find the source to this phenomenon...
What I do is filter the mappings in the SAM file and use only mappings with 1mismatch. This way I can be sure I'm using the mappings I want for the rest of the analysis.
But it still doesn't feel very good to use a program that has an unexpected (or un-understood) behavior...

Rachelly.

[nupurgupta]

Hi,
Actually I did figure something out for this one - see

http://sourceforge.net/mailarchive/forum.php?thread_name=CAKAActbbGTHJbwNQPu3TUGECgGuob0FUWTU1TF6WRVRqrdx7ng%40mail.gmail.com&forum_name=bio-bwa-help

Hope it helps!

[Rachelly]

Thanks for this note nupurgupta!
I actually see that this happens with mapping of single reads as well, and not only in PE as written in the above thread - I get mappings with more mismatches than what I've asked for...
I'm doing a filtration step after the mapping, to make sure I use only the mappings I want, but I don't understand why BWA allows to limit the amount of mismatches and then gives mappings with more mismatches..

I tried to add a reply to the thread you've mentioned, but wasn't able to.