How to align incomplete pairs and singled end reads?

Question

[loy_loy]

Hi everyone,

I converted bam files to fastq files to realign them with BWA. For some bam files, I get single end read files and files with incomplete pairs. I used the tool bam2fastq by biobambam2 which states single end reads as _s.fq.gz and incomplete pairs as _o1.fq.gz + _o2.fq.gz.

How do I align these files in the next steps? For some read groups i get all 3 outputs (also the "normal" output for complete read pairs which I usually use in the further analysis).

Thank you!

Best

Lynn

[lieven.sterck]

hard to say without knowing all the details.

Generally spoken one would drop all the single (and incomplete pairs) end data given there is enough paired end data left (eg. 90% correct paired end -> proceed with only PE)

[loy_loy]

Thank you for your reply. I thought I'd try to align the single reads to my reference as well and merge them later with the other aligned read groups that contain paired end reads.

[lieven.sterck]

a tempting thing to do indeed (we've all been there ;) ). In most cases that is an option, however it sometimes is a bit tricky to do (your specific case should be ok though at first sight).

Question remains on the other hand whether you need to do it (and thus if you would not better avoid it?). Are for some of the samples the read counts of the paired end data so low you feel you need to include the "single-end" as well?