mutiple mapped reads FLAG/TAG in bwa alignment data

[totalnew]

Hi, All


I am doing analysis on bwa alignment data. I have a question about multiple mapped reads, what I like to do is to identify/filter any pairs that one or two reads of the pair is/are multiple mapped. In any case that there are two equally likely alignments (both reads are mapped), the mapping quality is zero, and bwa will randomly pick one. That can be easily filtered out since the mapping quality of both reads is zero. On the other hand, if one read is uniquely mapped and the mate is mapped to multiple hits, that means unique mapped read has the good mapping quality, like 37, we can see this from XT tag, XT:AUnique, Repeat, or forced to map by smith-waterman).

I tried to filter out both reads (one read is with PE mapping quality higher than 0, the mate is 0). The mate read can be filtered out by mapping quality threshold 0. But I could not filter the first read because there is no clue to tell us whether the mate of this read is multiple mapped. I tried AM TAG, but AM is smaller single end mapping quality of two reads in a pair. It could be different from smaller pair end mapping quality of two reads. For example, this read

HS1_220:4:22:9461:127198 1107 1 28 25 100M = 2 -126 CCCTAACCCTACCCCCAACCCTAACCCTAACCCTACCCCTAACCCTAACCCTACCCCTAACCCTAACCCTAACCCTAACCCTACCCTAACCCTAACCCTA #########################?:9:6::>2;5BBB?7::=9-@=@A965'4??692AAA:=(8:<;6'75;9?'A5AB@DD@=>;BCCCC>>>=:B XT:A:U NM:i:5 SM:i:25 AM:i:0 X0:i:1 X1:i:0 XM:i:5 XO:i:0 XG:i:0 MD:Z:11A3T19A17A27A18

First read is uniquely mapped (XT:A:U), and its mate single end mapping quality is 0 (AM:i:0), so if I filter this read, I will make mistake since its mate paired-end mapping quality is 3. The mate is like below

HS1_220:4:22:9461:127198 1187 1 2 3 100M = 28 126 AACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCGAACCCTAACCCTAACC DDDDD?EDEEFGDGGGGFDGGFG?FDFDDFGGBGFGEEGEDFEDFEBAB==:==?FFFF?FAEE?################################### XT:A:R NM:i:1 SM:i:0 AM:i:0 X0:i:296 XM:i:1 XO:i:0 XG:i:0 MD:Z:83T16

So I wonder if anyone come cross this type of scenario, and how to deal with it, and I hope a new FLAG/TAG can be added in bwa bam files to specify the mate paired-end mapping quality.

thanks