Looking for MQ for non-variants, unsure why MMQ given as 0 for them.

Hello,

I'm doing quality control on outputs, both variant and non-variant, and I was hoping to have mapping quality for non-variants as well as variants but they are not given. I have produced my file using HaplotypeCaller and then produced a vcf from that file using GenotypeGVCFs with the parameter -ERC GVCF. I tried using BP_RESOLUTION but non-variants still did not possess MQ scores. In addition I am unclear as to why the MMQ is 0 for all non-variants as I did not change the minimum mapping quality score from 10 in the previous steps which I understand to be the default.

Thanks

Tagged:

Answers

  • bshifawbshifaw moonMember, Broadie, Moderator admin

    Hi @kingpg,

    Mind sharing the command you're using for Haplotypecaller and GenotypeGVCFs. Also a snippet of the vcf being produced.

    Thanks.

  • kingpgkingpg Member

    Sure thing @bshifaw,

    ./gatk HaplotypeCaller -I markdup_output.bam -R human_g1k_v37.fasta -O final.g.vcf -L filter.vcf -L wrong_positions.intervals -D filter.vcf -ERC GVCF

    ./gatk GenotypeGVCFs -O final.vcf -R human_g1k_v37.fasta -V final.g.vcf -all-sites True

    CHROM POS ID REF ALT QUAL FILTER INFO FORMAT N1_S1_L002_R1_001

    1 97079071 . G . 100.23 . DP=29 GT:AD:DP:RGQ 0/0:29,0:29:72
    1 97082391 . T . 124.23 . DP=32 GT:AD:DP:RGQ 0/0:32,0:32:96

    Is that enough?

    Thanks

  • kingpgkingpg Member

    I forgot to mention that I applied this afterwards to get the mentioned metrics.

    /gatk VariantAnnotator -O final_annotated.vcf -V final.vcf -RF MappingQualityReadFilter -A MappingQuality

  • bshifawbshifaw moonMember, Broadie, Moderator admin

    @kingpg
    That should be good, also what version of GATK are you using?

  • bshifawbshifaw moonMember, Broadie, Moderator admin

    Hi @kingpg

    I spoke with the dev team and here is their reply

    1. It is not possible to coerce HC to output a mapping quality for non-variant blocks.
    2. MMQ is median mapping quality and if there is no mapping quality data in the first place the median should end up being 0
Sign In or Register to comment.