How to filter paired reads with one is mapped and the other is unmapped

[Shenglin]

Hi everybody,

I want to filter the paired reads that one is mapped and the other is unmapped from bam file created by Bowtie2 with paired-end reads. Can it be done by samtools view?

Many thanks!

[dpryan]

samtools view -f 8 -F 4 -b foo.bam > foo.filtered.bam

That'll get you reads that are mapped with unmapped mates.

[Shenglin]

Thanks！I am new to samtools. The following is my analysis.

samtools view -f 8 -F 4 -b from.bam > mapped.bam
This is for the mapped reads with unmapped mates
Flagstat:
3390726 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
3390726 + 0 mapped (100.00%:-nan%)
3390726 + 0 paired in sequencing
1509156 + 0 read1
1881570 + 0 read2
0 + 0 properly paired (0.00%:-nan%)
0 + 0 with itself and mate mapped
3390726 + 0 singletons (100.00%:-nan%)
0 + 0 with mate mapped to a different chr
0 + 0 with mate mapped to a different chr (mapQ>=5)

samtools view -F 8 -f 4 -b from.bam > unmapped.bam
This is for the unmapped reads with mapped mates
Flagstat:
3390726 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
0 + 0 mapped (0.00%:-nan%)
3390726 + 0 paired in sequencing
1881570 + 0 read1
1509156 + 0 read2
0 + 0 properly paired (0.00%:-nan%)
0 + 0 with itself and mate mapped
0 + 0 singletons (0.00%:-nan%)
0 + 0 with mate mapped to a different chr
0 + 0 with mate mapped to a different chr (mapQ>=5)

So the values for the paired reads with one is mapped and the other is unmapped:
one read: mapped, the mate is unmapped
the other: unmapped, the mate is mapped

[dpryan]

-F 8 -f 4 will yield unmapped reads with mapped mates, but flagstat doesn't count mapped mates unless the read being looked at is also mapped (thus, the singleton metric with -f 8 -F 4).

[Shenglin]

Yes, many thanks!
I have another question: could samtools view be used to filter paired reads with upstream read is reverse stranded and downstream read is forward stranded (that means two reads with diverge direction)?

[Brian Bushnell]

Yes, many thanks!
I have another question: could samtools view be used to filter paired reads with upstream read is reverse stranded and downstream read is forward stranded (that means two reads with diverge direction)?

Those will be considered improper pairs, if the aligner was assuming normal Illumina fragment library orientation, so you just need to filter for proper pairs.

[dpryan]

Those will be considered improper pairs, if the aligner was assuming normal Illumina fragment library orientation, so you just need to filter for proper pairs.

Jein, to borrow a word from German. While looking for mapped reads lacking the properly-paired bit will likely get the alignments in question, you'll also get alignments distanced further apart than the aligner expected (or on different chromosomes, should any such alignments exist). The only reliable way to do this is to script things in a convenient language (I'd recommend python with pysam).