Question on mapping quality and uniquely mapped read

[azi]

Hi

I also use tophat to map RNA-seq data, and I have some questions, does the tag NH:i:1 work for pair-ended reads? Or it just concerns one mate read?

I mean, if I accept all alignments which have the tag NH:i:1, could I be sure that specific pairs which can be alined only one time, are all selected?

In other words, is it possible one mate has "NH:i:1", an the other mate has "NH:i:3"?

[ttnguyen]

Hope this is not too late to ask a question on this topic.
Is it the right way to extract unique-reads by just simply basing on the read name in SAM/BAM file (after filtering out reads having no reported alignments, i.e. samtools view -F4) ? It means a read is called unique-mapped if its name occurs exactly one time (for single-end). I asked this because I were afraid that different aligners use different ways to report mapping quality.

[nilshomer]

Mapping quality of zero usually means the mapping could not be unambiguously placed, which is likely in repeat regions.

[chenyao]

I figure out: using grep -w or awk.

Another question : lots of reads (mapping quality=0) are aligned to reapeted region in my RNA-seq data. what does this mean?

[chenyao]

What is the alignment score (AS tag) for those mapq = 0 reads? Can you post one record?

After checking the results, I found using grep "XH:i:1" I mistook the reads with "XH:i:1**", see the example:

2045_1360_1504_F5-BC 409 chr7 3155885 0 25M * 0 0 AAAAGGGGGAAGGGAAAGAGGGAGG 004DI:.""-,*"'/);LRYMDNCC NM:i:0 NH:i:10CC:Z:= CP:i:70303403 XS:A:+ HI:i:8


So what is the script which I can extract "XH:i:1" ?

[nilshomer]

What is the alignment score (AS tag) for those mapq = 0 reads? Can you post one record?

[chenyao]

Dear all.

I use tophat to map RNA-seq data. After getting results, I use two ways to filter the alignment. one is using grep "NH:i:1" to get uniquely mapping reads; the other is using samtools view -bq to get reliable reads. However, I found many uniquely mapped reads have mapping quality =0. How to explain it?
It's a single-end data.