BWA aligned pairs read1=XT:A:U, read2=XT:A:R

[jflowers]

Hello,

I've aligned Illumina paired end reads with BWA, filtered using samtools view -f 3 -F 12 ("proper pairs", both mates mapped).

I'm looking at the XT tags and see expected classes, where both mates are XT:A:U (both "unique") and XT:A:M + XT:A:U (one mate-rescued, one "unique").

However, I also find a class of paired reads where one read is XT:A:R (multiply-mapped) and the other is XT:A:U. This class of reads is a significant fraction of the total pairs (~10%) and the class persists after filtering on MAPQ > 20.

I anticipated that this class of reads would instead be classifed in the XT:A:M + XT:A:U (ie XT:A:R converted to XT:A:M), but apparently not.

Are the XT:A:R whose mate is XT:A:U mate-rescued, or something else?

Thanks.

[davetang]

I have exactly the same question and I haven't been able to find more detailed information on the XT:A:M tag from BWA.

I created a simple example (on my blog at http://davetang.org/muse/2014/07/24/...red-end-reads/; specifically the "RESCUING AMBIGUOUSLY MAPPED READS" section) trying to obtain the XT:A:M tag by aligning a uniquely- and multi-mapping pair but I got the XT:A:U and XT:A:R tags instead.