how to speed up SNP calling using bcftools

0

Entering edit mode

2.7 years ago

Chenfei Zheng ▴ 60

Hi, I was calling SNPs with bcftools pipeline as below:

bcftools mpileup --threads 20 -a AD,DP,SP -Ou -f ${REF} ${BAM}/*.bam | bcftools call --threads 20 -f GQ,GP -mO z -o ./output.vcf.gz

But the option --threads didn't work and this calling has taken about two weeks to output a 71G vcf file.

According to the instruction of bcftools only work for compression of output file when the output type is b or z. I've changed the output type to -Ob, but multiple threads still didn't work either.

So, how can I speed up this calling process?

Thank you

bcftools • 1.7k views

ADD COMMENT • link updated 2.7 years ago by Ram 43k • written 2.7 years ago by Chenfei Zheng ▴ 60

0

Entering edit mode

So, how can I speed up this calling process?

parallelize by regions using --regions-file

ADD REPLY • link 2.7 years ago by Pierre Lindenbaum 161k

0

Entering edit mode

Thank you for advice. Can I parallelize by individual and then merge vcf.gz files together?

ADD REPLY • link 2.7 years ago by Chenfei Zheng ▴ 60

0

Entering edit mode

Can I parallelize by individual

it's a job for gatk HaplotypeCaller in GVCF mode.

ADD REPLY • link 2.7 years ago by Pierre Lindenbaum 161k

0

Entering edit mode

The --threads argument to bcftools call is useless IMO - it is the number of threads used for compression, not used across the board when the output is set to be compressed.

ADD REPLY • link 2.7 years ago by Ram 43k

Login before adding your answer.