To celebrate the release of GATK 4.0, we are giving away free credits for running the GATK4 Best Practices pipelines in FireCloud, our secure online analysis portal. It’s first come first serve, so sign up now to claim your free credits worth $250. Sponsored by Google Cloud. Learn more at https://software.broadinstitute.org/firecloud/documentation/freecredits

VQSR failed with "No data found" with whole genome variant calls, but succeeded with filtered set

Hi,

I ran VQSR on the SNP calls from a WGS sample mapped to b37+decoy reference, it failed at the training step with error message "No data found". I then removed the SNP calls on the small contigs (unmapped/random/decoy) and only retain the ones mapped to the major chromosomes, and passed this new set to VQSR training with the same parameters again. It was successful.

From what I understood (but I may be misunderstanding) from the VQSR documentation, the gaussian mixture model construction should not rely on chromosomal locations, but only the variant annotations. If the whole genome job failed with "no data found", either no enough bad variants or good variants, I'm wondering how could it then find enough data with the filtered set?

For comparison I ran VQSR for another WGS sample which succeeds on the training both unfiltered or filtered. It seems the FILTER column in the final results will have lots of differences between filtered and unfiltered, which can be expected because the variants pool has been changed. But I'm also wondering are both results OK to use for downstream analysis?

Thank you very much for providing any information/comment.

Tagged:

Comments

  • BTW, I'm using GenomeAnalysisTK-3.4-46.

  • SheilaSheila Broad InstituteMember, Broadie, Moderator

    @jianl
    Hi,

    Can you please post the exact command you ran?

    Thanks,
    Sheila

  • Here's my script to run on the filtered set (excluding small contigs). To run without filtering, I just remove the "-XL" option.

    EXCLUDE="GL000207.1 -XL GL000226.1 -XL GL000229.1 -XL GL000231.1 -XL GL000210.1 -XL GL000239.1 -XL GL000235.1 -XL GL000201.1 -XL GL000247.1 -XL GL000245.1 -XL GL000197.1 -XL GL000203.1 -XL GL000246.1 -XL GL000249.1 -XL GL000196.1 -XL GL000248.1 -XL GL000244.1 -XL GL000238.1 -XL GL000202.1 -XL GL000234.1 -XL GL000232.1 -XL GL000206.1 -XL GL000240.1 -XL GL000236.1 -XL GL000241.1 -XL GL000243.1 -XL GL000242.1 -XL GL000230.1 -XL GL000237.1 -XL GL000233.1 -XL GL000204.1 -XL GL000198.1 -XL GL000208.1 -XL GL000191.1 -XL GL000227.1 -XL GL000228.1 -XL GL000214.1 -XL GL000221.1 -XL GL000209.1 -XL GL000218.1 -XL GL000220.1 -XL GL000213.1 -XL GL000211.1 -XL GL000199.1 -XL GL000217.1 -XL GL000216.1 -XL GL000215.1 -XL GL000205.1 -XL GL000219.1 -XL GL000224.1 -XL GL000223.1 -XL GL000195.1 -XL GL000212.1 -XL GL000222.1 -XL GL000200.1 -XL GL000193.1 -XL GL000194.1 -XL GL000225.1 -XL GL000192.1 -XL NC_007605 -XL hs37d5"

    java -Xmx30g -jar $gatk -T VariantRecalibrator -R $ref -input $1.vcf -resource:dbsnp,known=true,training=false,truth=false,prior=2.0 ${resourceDir}/dbsnp_138.b37.vcf -resource:hapmap,known=false,training=true,truth=true,prior=15.0 ${resourceDir}/hapmap_3.3.b37.vcf -resource:omni,known=false,training=true,truth=true,prior=12.0 ${resourceDir}/1000G_omni2.5.b37.vcf -resource:1000G,known=false,training=true,truth=false,prior=10.0 ${resourceDir}/1000G_phase1.snps.high_confidence.b37.vcf -an QD -an MQ -an MQRankSum -an ReadPosRankSum -an FS -an DP -mode SNP -recalFile $1.recal -tranchesFile $1.tranches -nt 4 -XL ${EXCLUDE}

    java -Xmx30g -jar $gatk -T ApplyRecalibration -R $ref -input $1.vcf -tranchesFile $1.tranches -recalFile $1.recal -o $1.recalibrated.vcf --ts_filter_level 99.5 -mode SNP -XL ${EXCLUDE}

  • SheilaSheila Broad InstituteMember, Broadie, Moderator

    @jianl
    Hi,

    Just to be sure, can you try using the latest version of GATK (3.6) and let us know if the issue persists?

    Thanks,
    Sheila

Sign In or Register to comment.