strand-specific bam files

[sheremey]

Hello all,

I know this must be a very simple and common task, sorry if I've overlooked an existing recommendation while searching the forum.

Given: BAM file with aligned paired-end reads
Needed: split it into 2 strand-specific BAM files (first read in each aligned pair should correspond to F strand of the reference in the first BAM file and to the R strand of the reference in the second BAM file).

Thanks a lot!

[maubp]

Look at the samtools view -f and -F options, you could do this by filtering on the FLAG 0x10 values.

(If you have paired data things are more interesting...)

[sheremey]

Look at the samtools view -f and -F options, you could do this by filtering on the FLAG 0x10 values.

(If you have paired data things are more interesting...)

I believe the FLAG relates to an individual read, right? My first BAM file needs to have first reads aligned to F and second reads to R of the reference. I cannot believe there is no standard solution for this task (that is supposed to be popular in downstream analysis) for paired-end data that is overwhelmingly popular now... Is this indeed the case?

[maubp]

I don't know why you would want to do (what popular downstream analysis exactly?), but you could do this with 'samtools view' and the FLAG filters.

Before spending any effort on this, please clarify if you want file one to have reads like this:

       1 ---> 
=========================
               <--- 2

And file two to have:

       2 ---> 
=========================
               <--- 1

[sheremey]

I don't know why you would want to do (what popular downstream analysis exactly?), but you could do this with 'samtools view' and the FLAG filters.

Before spending any effort on this, please clarify if you want file one to have reads like this:

       1 ---> 
=========================
               <--- 2

And file two to have:

       2 ---> 
=========================
               <--- 1

Thank you,

Yes, that's exactly the desired outcome.
As to downstream analysis, I am talking about investigating on antisense transcription phenomena (given a strand-specific library), which is going to be increasingly popular down the road. Thanks a lot!

[swbarnes2]

Put the reads with flags of 99 (read 1, mapped forward, mate mapped reverse, properly paired) and 147(read 2, mapped reverse, properly paired) in file 1, and the reads with flags 83 and 163 in file 2.

[maubp]

I concur file one needs:

0x1 - paired
0x2 - properly paired
0x20 - partner on reverse strand
0x40 - read one
FLAGs 0x1 + 0x2 + 0x20 + 0x40 = 0x63 = 99 in decimal

And file one also needs:

0x1 - paired
0x2 - properly paired
0x10 - on reverse strand
0x80 - read two
FLAGs 0x1 + 0x2 + 0x10 + 0x80 = 0x93 = 147 in decimal

So I'd use 'samtools view' twice to get those reads, then 'samtools merge' to combine them into file one.

Likewise for file two:

0x1 - paired
0x2 - properly paired
0x10 - reverse strand
0x40 - read one
FLAGs 0x1 + 0x2 + 0x10 + 0x40 = 0x53 = 83 in decimal

And file two also needs:

0x1 - paired
0x2 - properly paired
0x30 - partner on reverse strand
0x80 - read two
FLAGs 0x1 + 0x2 + 0x20 + 0x80 = 0xA3 = 163 in decimal

[sheremey]

I concur file one needs:

0x1 - paired
0x2 - properly paired
0x20 - partner on reverse strand
0x40 - read one
FLAGs 0x1 + 0x2 + 0x20 + 0x40 = 0x63 = 99 in decimal

And file one also needs:

0x1 - paired
0x2 - properly paired
0x10 - on reverse strand
0x80 - read two
FLAGs 0x1 + 0x2 + 0x10 + 0x80 = 0x93 = 147 in decimal

So I'd use 'samtools view' twice to get those reads, then 'samtools merge' to combine them into file one.

Likewise for file two:

0x1 - paired
0x2 - properly paired
0x10 - reverse strand
0x40 - read one
FLAGs 0x1 + 0x2 + 0x10 + 0x40 = 0x53 = 83 in decimal

And file two also needs:

0x1 - paired
0x2 - properly paired
0x30 - partner on reverse strand
0x80 - read two
FLAGs 0x1 + 0x2 + 0x20 + 0x80 = 0xA3 = 163 in decimal

Thank you! That's outstanding!