[GenoMax]

Look into freebayes. There is samtools/bcftools based option as well.

[cmccabe]

sorted bam and bbmap

Is there an option for bbmap that will result in a sorted bam? I know adding "out=mapped.bam" this will produce a bam but it appears to be unsorted:

samtools view -H /home/cmccabe/IDP/mapped_unmatched_adapter_removed_quality_trimmed_NA12878-NextSeq-S1_R1_and_R2.bam
@HD	VN:1.4	SO:unsorted

Thank you .

[GenoMax]

No. You have to do some work

$ samtools sort your_file.bam your_file_sorted

[cmccabe]

just checking, thank you .

bbmap is amazingly fast and accurate . Great tool.

[Brian Bushnell]

No. You have to do some work

$ samtools sort your_file.bam your_file_sorted

Of course, I do my best to prevent needing people to do work, so...

bbmap.sh in=reads.fq out=mapped.bam bs=bs.sh; sh bs.sh

That will write a script "bs.sh" that contains the command to make a sorted, indexed bam file, then execute it after BBMap is done. It saves... well, at least a couple of keystrokes

[GenoMax]

Hmm. I should start a new thread for "Easter eggs" in BBMap

[westerman]

Couldn't you create the shell script and then execute it within BBMap?

[Brian Bushnell]

Couldn't you create the shell script and then execute it within BBMap?

Unfortunately, not easily, due to memory constraints. BBMap's shell script already grabs all available memory by default. It can launch a samtools process for conversion to bam, which is a low-memory streaming process; but sorting uses a lot of memory (to do it most efficiently), so it might run out of memory. Java cannot reduce its virtual memory allocation after starting, and won't share that memory with other programs.

It would work fine on most individual computers since physical memory limits would never be exceeded, but job schedulers often have strict limits on virtual memory and will kill jobs that exceed it. So, it would reduce stability (and typically fail) on clusters unless I did parsing in the shellscript to launch a new process after BBMap terminates, or the user launched BBMap with a custom -Xmx flag. I try to keep the shellscripts as simple as possible. Anyway, there are workarounds, but they all add complexity due to memory issues.