extremely large vcf file generated by GATK

[slowsmile]

Dear All
I am new to snp analysis and am recently working on GATK pipeline to identify snps on the whole genome data.

I went through the stage I raw data processing protocol recommended by GATK pipeline and generated an aligned, duplicate marked, and recalibrated BAM file for one of my samples.

I tried to run the Unified genotyper with the following code (for both snps and indels)

java -Xmx4g -jar GenomeAnalysisTK.jar \
-l INFO \
-R /media/edrive/RNAseq/ucsc.hg19.fasta \
-T UnifiedGenotyper \
-I /media/edrive/RNAseq/LID46437/bwa/reads.recal.bam \
-o /media/edrive/RNAseq/LID46437/bwa/output.vcf \
--genotype_likelihoods_model BOTH \
--output_mode EMIT_ALL_CONFIDENT_SITES

exit

The code has been running for more than 20 hours yet has not completed. The in put bam file (reads.recal.bam) is only 11GB but as I checked the size of output vcf file, it already reached 60GB.

I just want to ask whether this is normal to take such a long time to run Unified genotyper and generates an excessively large vcf file?

Can anyone with experiences tell me: what is the size of your output vcf file when you work on the whole human genome? And how long does it take to run Unified genotyper usually?

Thanks a lot

[shawpa]

When I ran unified genotyper for the first time I was a bit confused by the "output" mode. You selected emit_all_confident_sites. Another researcher told me that he believed that choosing this as your output would output any site that is listed as variable in the population but not necessarily your sample. So you could get sites where your sample is homozygous for the reference rather than it outputing only the variants. I ran unified genotyper under default mode on a 80GB file and I think it only took like 8 hours at the most (over 24 processors).

I wouldn't take my word for it, but maybe someone else will post.

[kissthefuture]

Dear All
I am new to snp analysis and am recently working on GATK pipeline to identify snps on the whole genome data.

I went through the stage I raw data processing protocol recommended by GATK pipeline and generated an aligned, duplicate marked, and recalibrated BAM file for one of my samples.

I tried to run the Unified genotyper with the following code (for both snps and indels)



The code has been running for more than 20 hours yet has not completed. The in put bam file (reads.recal.bam) is only 11GB but as I checked the size of output vcf file, it already reached 60GB.

I just want to ask whether this is normal to take such a long time to run Unified genotyper and generates an excessively large vcf file?

Can anyone with experiences tell me: what is the size of your output vcf file when you work on the whole human genome? And how long does it take to run Unified genotyper usually?

Thanks a lot

You may chose --output_mode EMIT_VARIANTS_ONLY instead.