Running time in different callers

Question

[zizigolu]

Hi,

I have some .bam files from WGS their average size is 70000000 KB. I called SNV by Strelka but output .vcf files are not compatible with some downstream programs so I tried Varscan, mutect2 and muse but I noticed even with 16 cpu calling mutations never finished finally session killed. even I tried mutect2 on one chromosome by L option but again takes an ages. I am wondering how people use these tools when they never gonna finished in calling mutations

My code for mutect2

java -jar /home/local/software/GATK/3.7/source/GenomeAnalysisTK.jar -T MuTect2 -R /temp/hgig/fi1d18/hs37d5.fa -I:tumor /scratch/fi1d18/example_results/1631_WTSI-COLO_075_1pre/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_1pre.dupmarked.bam -I:normal /scratch/fi1d18/example_results/1631_WTSI-COLO_075_b/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_b.dupmarked.bam -L 1  -o out.vcf

Varscan

/home/local/software/GATK/3.7/source/varscan somatic <(samtools mpileup --no-BAQ -f /temp/hgig/fi1d18/hs37d5.fa   /scratch/fi1d18/example_results/1631_WTSI-COLO_075_b/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_b.dupmarked.bamm /scratch/fi1d18/example_results/1631_WTSI-COLO_075_1pre/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_1pre.dupmarked.bam) /wgs --mpileup 1 --output-vcf

I selected these options on our cluster

-l walltime=36:00:00 -l nodes=1:ppn=16

But jobs never done and killed after 36 hours

What people do on such WGS files?

[h.mon]

-l nodes=1:ppn=16

I don't see, in the varscan and mutect2 command-lines, 16 threads being assigned to the programs. If you reserve 16 threads with the resource manager but don't assign them to the software with the appropriate command-line argument, most programs will use one thread by default.

[Nicolas Rosewick]

You should use gatk 4 . What is the amount of RAM you have ?

[zizigolu]

Sorry, I am also trying GATK 4.0 but is the same. How I know the amount of RAM I have? How I can use max RAM?

[Ram]

You should check with your cluster on RAM and wall-time limits. For GATK, you should use at least ~16G RAM for good performance.