Variant calling parallelization

[Anonymous]

Hi NGSEP devs,
I was wondering if the multiple variant caller can be parallelized? I found the option (choosing the number of threads) in the alignement method, but not in the variant caller.
Best regards

[Jorge Duitama]

Hi

Thanks for your interest in NGSEP. Unfortunately we are still working on the paralellization of the variant caller. We hope to make progress in this direction in the next few months.

Best regards

Jorge

[Anonymous]

Thanks Jorge.

I've seen on the web that one can at least treat chromosomes in parallel, for instance, using gnu parallel:

parallel 'bcftools mpileup -Ou -f ref.fa -r {} in.bam | bcftools call -m -v -Oz -o {}.vcf.gz' ::: {1..22} X Y

(from https://github.com/samtools/bcftools/issues/949)
Is it the way you are thinking in?

Best regards

[Jorge Duitama]

Hi

Now that you mention it, parallel can be a way to run in parallel chromosomes. The command in our case would be something like this:

parallel 'java -Xmx8g -jar /path/to/NGSEPcore_4.2.0.jar MultisampleVariantsDetector -r ref.fa -maxBaseQS 30 -minQuality 40 -querySeq {} -o {}.vcf in.bam' ::: {1..22} X Y

We plan to implement parallelization within the software to distribute the genome by regions instead of chromosomes, which should work better if you have more than 24 threads. By now, parallel can be a good solution.

Thanks

[Anonymous]

Thanks Jorge. I will try with the command you suggested. Have a nice day.