In Gkno'S Fastq-Vcf Pipeline, Can I Write To Disk An Intermediary Bam Aligment Currently Sent To The Stream?

Question

[Carlos Borroto]

Hi,

I'm using gkno's 'fastq-vcf' pipeline as a way to run a serial of steps for variant calling. gkno allows the use of bash streams to pipe data between tools without ever writing to disk.

I was wondering if there is away to keep the BAM alignment coming out of "left-align-indels" step in this pipeline. I can see these three steps are writing to the stream: "filter-bam", "realign-gaps", "left-align-indels"

And "variant-call"(freebayes) is the receptor of the final stream. I would like to instead save the intermediary BAM file for further visualization and analysis. If this is possible, do you think this change would cause a big performance hit?

I did look at bamleftalign's(tool used for "left-align-indels" step), it seems like the only option is to write to the stream. Is that right?

Answer 1

[alistairnward]

I have modified how output streams are handled and included an additional pipeline - fastq-bam. This pipeline is identical to the fastq-vcf pipeline except that it terminates after left-aligning indels and produces a BAM file as output. Please update to the most recent version of GKNO and these changes should be available.

Answer 2

[Alex Reynolds]

Perhaps you could insert an instance of the UNIX tee command, piping streamed intermediate data to a named pipe that you can process separately. As this is a bash or other shell script, you should be able to read the script and insert the tee statement where you need it.

[Carlos Borroto]

That's a good suggestion. I might have to integrate 'tee' as a tool in gkno. If I do, that could meant I can easily grab the stream coming out of tools which currently only write to the stream.

Thanks, I will explore this option.