We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

genotypeFilterExpression outputs

I was using t-T SelectVariants with the the genotypeFilterExpression with DP<11. In the output some of the samples have show 'pass" or 'low dp", some doesn't, and the ./. also shown "pass". Is there anything wrong?


  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Hi there,

    Can you post a few of those records here? It's difficult to evaluate without seeing the records themselves.

  • yg1yg1 Member
    edited November 2013

    Sure, for example I got

    Locus_74797_Transcript_1/1_Confidence_1.000_Length_316 165 . C A 595.58 PASS AC=2;AF=1.00;AN=2;DP=15;FS=0.000;MLEAC=2;MLEAF=1.00;MQ=37.00;MQ0=0;QD=26.05 GT:AD:DP:GQ:PL ./. ./. ./. ./. ./. ./. ./. 1/1:0,15:15:45:620,45,0 ./. ./.

    and I also got

    Locus_75584_Transcript_1/1_Confidence_1.000_Length_418  120 .   A   T   420.55  PASS    AC=11;AF=0.917;AN=12;BaseQRankSum=0.869;ClippingRankSum=1.448;DP=12;FS=4.771;MLEAC=11;MLEAF=0.917;MQ=37.00;MQ0=0;MQRankSum=0.869;QD=32.58;ReadPosRankSum=1.448  GT:AD:DP:FT:GQ:PL   1/1:0,3:3:Low DP:9:123,9,0  1/1:0,2:2:Low DP:6:84,6,0   ./.:.:.:PASS    ./.:.:.:PASS    ./.:.:.:PASS    1/1:0,1:1:Low DP:3:42,3,0   1/1:0,1:1:Low DP:3:42,3,0   0/1:1,2:3:Low DP:32:75,0,32 1/1:0,2:2:Low DP:6:84,6,0   ./.:.:.:PASS

    So I the same output some of them were filtered with "low dp" (2nd examples) and some were not (1st example).

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Are these two records from the same file? What was your command line?

  • yg1yg1 Member
    edited November 2013

    These two records are fom the same file, my command is

    java -Xmx20g -jar /opt/software/GATK/2.7.2/GenomeAnalysisTK.jar \
      -R Axillaris_final_transcripts.fasta \
      -T VariantFiltration \
      -o SE10_Filter_SNPOnly_haplotypecaller.vcf \
      --variant SE10_rawSNPs_Only_Filter_haplotypecaller.vcf \
      --clusterWindowSize 50 \
      --filterExpression "MQ0 >= 4 && ((MQ0 / (1.0 * DP)) > 0.1)" \
      --filterName "HARD_TO_VALIDATE" \
      --filterExpression "SB >= -1.0" \
      --filterName "StrandBiasFilter" \
      --filterExpression " QD < 5.0 " \
      --filterName " QD Filter "  \
      --filterExpression "HRun >5" \
      --filterName "Hrun Filter" \
      --filterExpression "QUAL<10" \
      --filterName "Low QUAL" \
      --filterExpression "DP<10" \
      --filterName "DP LOW ALL" \
      --genotypeFilterExpression "DP<11 " \
      --genotypeFilterName "Low DP" 

    then I use the below command to select SNPs without and filtering, and that was the results I got.

    java -Xmx20g -jar /opt/software/GATK/2.7.2/GenomeAnalysisTK.jar  \
      -R Axillaris_final_transcripts.fasta \
      -T SelectVariants \
      --variant SE10_Filter_SNPOnly_haplotypecaller.vcf  \
      -o SE10_Filtereed.vcf  \
    Post edited by Geraldine_VdAuwera on
  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Oh I see. What you're trying to do is not currently possible. SelectVariants looks only at site-level filter status, it can't select based on genotype-level filter status.

  • yg1yg1 Member

    Sorry, I didn't make it clear. I'm not trying to make the genotype-level selection. When I looked at the results from the first file generated, I found some of them have "LOW dp" and "pass", some of then didn't have, and the ./. was also labeled as "pass" , just like the two lines shown above.

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    OK, so you're worried because it looks like the genotype filters weren't applied at all to some of your variants (e.g. the first example), correct? I agree it's strange that the first site you posted doesn't have the genotype-level FT field at all. If you upload a snippet of your VCF I'll take a closer look. I'll need a copy of your reference as well.

Sign In or Register to comment.