Generating dicordant and splitter bam file about speedseq HOT 6 CLOSED

Hi Xiaotong,

Yes, that's correct. If you do not use SpeedSeq for alignment, you'll have to generate discordant and split reads. We already have a tool for doing this, called Extract SV Reads. You can find the software and documentation here: https://github.com/hall-lab/extract_sv_reads.

Best,
Colby

from speedseq.

What aligner were you using? We have only tested samblaster's splitter/discordant extraction on BWA-MEM aligned files (using the -M flag to mark shorter split reads as secondary).

Also, is your BAM file sorted by read ID?

from speedseq.

We are using bwa mem (0.7.5a-r405) , yes we do mark them using -M and yes the bam file was read name sorted for samblaster to run. These is the example command:
bwa mem -t 4 -PM

Thanks for your quick reply.

from speedseq.

Interesting, I'm not sure why that wouldn't work. Could you send me the first 10000 or so lines from your BAM to cc2qe at virginia dot edu? I probably won't have time to look at it until tomorrow at the earliest though.

In the meantime, we have this script that may work on your file:
https://github.com/hall-lab/svtools/blob/master/extractSplitReads_BwaMem

from speedseq.

BAM file lacked secondary reads, potentially due to different behavior of BWA MEM prior to v0.7.7.
closing.

from speedseq.

Hi @cc2qe ,

So if the BAM is aligned by bwa aln and not sorted by QNAME, then the internal default spl extraction of speedseq sv won't work, right?

I noticed the discordant reads were generated properly but not split reads. In contrast, if I explicitly run speedseq realign on my BAMs first, then the split reads extraction is successful, but this step also does the entire BAM which is a big waste of time when we are only interested in the reads supporting potential SVs.

Do you have the plan to make the default behavior of speedseq sv to be compatible with bwa aln BAMs?

Thanks,
Xiaotong

from speedseq.