This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!
Genotypegvcfs 3 Week Estimated Run Time for WGS
I am currently running Genotypegvcfs on 86 human WGS using GATK 3.7 and I am experiencing an estimated run time of 3+ weeks. My code is:
java -Xmx120G -Djava.io.tmpdir=/mnt/data_schen/travis/WGS-86/tmp -jar /mnt/data2/travis/bin/gatk_3.7\GenomeAnalysisTK.jar \
-T GenotypeGVCFs \
-R /mnt/data2/travis/REFERENCE/human_g1k_v37_decoy.fasta \
--variant WGS_86.list \
I am using GATK 3.7 to keep my pipeline consistent as I plan on combining these 86 with another 101 WGS samples that were run last year. HTC was used to generate gvcfs for each of the 86 individuals.
When run in multi-thread mode the application crashed with the "Code exception" error. This is very unfortunate as it had an estimated run time of 6 days.
I have read that you suggest running Genotypegvcfs in parallel by separating gvcfs based on chromosome. Do I need to:
- Separate every gvcf of every sample into 22 chromosomes + sex chromosomes (i.e. 187x24) with SelectVariants
- Combine the chromosomes using Combinegvcfs
- Run each chromosome separately through Genotypegvcfs
- Then combine the chromosomes using CatVariants
I am uncertain I fully understand the current recommendations and am looking for advice before moving forward. Are there any alternative methods I may consider?