GATK combine gvcf superslow

Question

[Picasa]

Hi,

I am working on a classic variant calling on a non model organism.

I have 250 samples and followed the GATK best practice.

I have produced 250 g.vcf with HaplotypeCaller, now the next step is to combine those g.vcf and produce a .vcf (with GenotypeGVCFs) either:

A) Solution A: using GenomicsDBImport

B) Solution B: using CombineGVCFs

But those methods are super slow.

I am wondering if it is possible to produce one vcf per g.vcf with GenotypeGVCFs (quite fast) and then combine the 250 vcf with an another program ?

Does it produce the same result ? Thanks.

[Ram]

CatVariants should help with the combine part of your workflow, I think.

[Pierre Lindenbaum]

split by chromosome.

[Picasa]

I am working on a non model organism, with a genome that have been assembled. Unfortunately, this is quite fragmented. Is is still worth do it ?

[Pierre Lindenbaum]

if there are 1000 contigs and you can run 1000 CombineGVCFs jobs in parallel, then it will be 1000 times faster..

[Nicolas Rosewick]

Hi Pierre. Do you split by chromosome for combineGVCFs or GenomicsDBimport ?

[Pierre Lindenbaum]

yes .

[Nicolas Rosewick]

should also work for genomicsDBimport though :)