VQSR dataset preparation

Answered

Follow

sabrina legoueix

• June 10, 2021 14:47

Hi,

My information are :

a) GATK version used: gatk-4.2.0.0
b) Exact command used:

gatk -Xmx40g VariantRecalibrator \
-R $genome \
-V cohort.vcf.gz \
-an QD -an MQRankSum -an ReadPosRankSum -an FS -an MQ -an SOR -an DP \
-mode SNP \
--resource:test1,known=false,training=true,truth=true,prior=10 test1.vcf.gz \
--resource:CMJ,known=false,training=true,truth=true,prior=10 CMJ.vcf.gz \
--resource:dbsnp,known=true,training=false,truth=false,prior=2 dbsnp.vcf.gz \
-O cohort_snps.recal \
--tranches-file cohort_snps.tranches \
--rscript-file output.plots.R

gatk -Xmx40g VariantRecalibrator \
-R $genome \
-V cohort.vcf.gz \
-an QD -an MQRankSum -an ReadPosRankSum -an FS -an MQ -an SOR -an DP \
-mode INDEL\
--resource:test1,known=false,training=true,truth=true,prior=10 test1.vcf.gz \
--resource:CMJ,known=false,training=true,truth=true,prior=10 CMJ.vcf.gz \
--resource:dbsnp,known=true,training=false,truth=false,prior=2 dbsnp.vcf.gz \
-O cohort_snps.recal \
--tranches-file cohort_snps.tranches \
--rscript-file output.plots.R

As I have large datasets of confident variants, I do not need to use hard filtering. I am using vqsr directly.

If not an error, choose a category for your question(REQUIRED):

a) How do I prepare my training datasets for SNP and INDELS:

- do I need to create training datasets with INDELS only for the mode INDELS?

- do I need to create training datasets with SNPs only for mode SNPs?

Same question for the query cohort.vcf.gz? Do I need to separate INDELs and SNPs to run each mode separatly?

I have tried the run mode SNPs with all SNPs and INDELs for query and datasets and I do no t have an error, the snp.recalibrated.vcf.gz is produced. But I am wondering if I used the right method.

Thank you very much in advance for your answer.

Best regards,

Sabrina

 

0

• 

  Genevieve Brandt (she/her)

  • June 17, 2021 21:46

  Hi Sabrina,

  Please read these articles regarding how to run VQSR:

  • https://gatk.broadinstitute.org/hc/en-us/articles/360035531112--How-to-Filter-variants-either-with-VQSR-or-by-hard-filtering
  • https://gatk.broadinstitute.org/hc/en-us/articles/360035531612-Variant-Quality-Score-Recalibration-VQSR-

  Let me know if these do not answer your questions.

  Best,

  Genevieve

  0

  Comment actions Permalink
• 

  Genevieve Brandt (she/her)

  • July 09, 2021 20:47

  sabrina legoueix we have released a new article that covers your questions even more in depth: https://gatk.broadinstitute.org/hc/en-us/articles/4402736812443-Which-training-sets-arguments-should-I-use-for-running-VQSR-

  0

  Comment actions Permalink

Please sign in to leave a comment.