BWA pe alignment, sampe, proper pair

[kevlim83]

Hi,

I am trying to align a set of paired-end reads using bwa. I understand that the flags tell us whether a pair is mapped and proper. i.e. one on the forward strand and one on the reverse strand, and also within insert estimate.

Now the problem is that most of my flags are 97,145,81,161.

This really means one is mapped to forward and the other to reverse. I also manually checked insert size is within 5 stdev from the mean (according to the bwa output, the maximum inferred insert is 5.88 from mean, so i assume anything within 5.88 stdev is a concordant read). This means the pair should be proper? But why is the flag=2 not set? i.e. I expected to get flags 99,147,83,163 instead of 97,145,81,161.

Here is an example of my alignment results:
>test1 97 chr2 1941945 37 27M = 1940790 -1128 AAAAAAAAGCCCCGCCGACATTTCCAC ~~~~~~~~~~~~~~~~~~~~~~~~~~~ XT:A:U NM:i:1 SM:i:37 AM:i:37 X0:i:1 X1:i:0 XM:i:1 XO:i:0 XG:i:0 MD:Z:0G26
>test1 145 chr2 1940790 37 27M = 1941945 1128 CTGCAGCAGCACGTTCTGCATTTCCTT ~~~~~~~~~~~~~~~~~~~~~~~~~~~ XT:A:U NM:i:0 SM:i:37 AM:i:37 X0:i:1 X1:i:0 XM:i:0 XO:i:0 XG:i:0 MD:Z:27

The mean insert size is 1250 and stdev is 46, maximum inferred insert is 5.88 stdev from mean (as reported by bwa). So, 1128 should be within the range for concordant.

Does anyone knows why an alignment like this is not considered proper? i.e. why is flag=2 not set?

Regards,
Kevin

[kevlim83]

Anyone with ideas? Please help. Thanks!

[valei]

You are right.
I think the problem is the follow: If the first read (with flag 97) refers to forward strand, the second read must be downstream. It's not possible that the second read is at position -1128.

[swbarnes2]

Your reads are running away from each other, bwa apparently only calls them proper if they run into each other, like normal paired-end reads do.

bwa wants this:

---> <---

Your mate-pair reads are doing this:

<--- --->