Missing variants with Haplotype Caller runs using multiple threads

marcelcaraciolomarcelcaraciolo BrazilMember
edited October 2015 in Ask the GATK team

Hi GATK team,

First I'd like to thank you guys for the tools that you're making available for the community!

The problem is that I have run my sample using Haplotype Caller and I faced some missing variants when I ran the HaplotypeCaller running with nct.

How I figured out ?

2) When I ran with the command (with nct deactivated):

java -Xmx10g -jar  GenomeAnalysisTK.jar -R ucsc.hg19.fasta   -I 165019-0-LAOM-N10_26_001_L001_1.realigned.recal.bam  --dbsnp dbsnp_138.hg19.vcf    -T HaplotypeCaller -stand_emit_conf 30.0  -stand_call_conf 30.0  -dcov 5000 --genotyping_mode DISCOVERY -A FisherStrand -A AlleleBalance -A BaseCounts -A StrandOddsRatio -A StrandBiasBySample  --max_alternate_alleles 3 -o .165019-0-LAOM-N10_26_001_L001_1.realigned.recal.gatk.high.vcf -L NUTRI.list

chr12 48239835 rs1544410 C T 6129.77 . AC=1;AF=0.500;AN=2;BaseQRankSum=11.887;DB;DP=509;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQ0=0;MQRankSu
m=1.141;QD=12.04;ReadPosRankSum=0.418;SOR=0.640 GT:AD:GQ:PL:SB 0/1:261,247:99:6158,0,6273:261,0,247,0

1) When I ran with the command below (with nct activated) :

java -Xmx10g -jar  GenomeAnalysisTK.jar -R ucsc.hg19.fasta  -nct 8  -I 165019-0-LAOM-N10_26_001_L001_1.realigned.recal.bam  --dbsnp dbsnp_138.hg19.vcf    -T HaplotypeCaller -stand_emit_conf 30.0  -stand_call_conf 30.0  -dcov 5000 --genotyping_mode DISCOVERY -A FisherStrand -A AlleleBalance -A BaseCounts -A StrandOddsRatio -A StrandBiasBySample  --max_alternate_alleles 3 -o .165019-0-LAOM-N10_26_001_L001_1.realigned.recal.gatk.high.vcf -L NUTRI.list

The variant above is missing from my vcf (it's not called).

Checked the flags: DP = 509 (good depth); QD 12.04 > 2.0;

I have ran with the bamout option to see the variant with HaplotypeCaller and it shows the variant as you can see at the figure below.
image

This is the original bam as INPUT at my variant caller

image

My HaplotypeCaller is running at version 3.3.

PS: I have seen some threads at the forum about missing variants running with nct, is is correct?

Answers

  • marcelcaraciolomarcelcaraciolo BrazilMember

    Additional information:

    I ran with UnifiedGenotyper with the command below:

    java -Xmx10g -jar GenomeAnalysisTK.jar -R ucsc.hg19.fasta    -I 165019-0-LAOM-N10_26/alignments/165019-0-LAOM-N10_26_001_L001_1.realigned.recal.bam  --dbsnp dbsnp_138.hg19.vcf    -T UnifiedGenotyper -stand_emit_conf 10.0  -stand_call_conf 30.0  -dcov 5000 -glm BOTH  --max_alternate_alleles 3 -o 165019-0-LAOM-N10_26_001_L001_1.realigned.recal.unified.vcf -L NUTRI.list 
    

    And I got the same SNP .

    chr12 48239835 rs1544410 C T 41707.77 . AC=1;AF=0.500;AN=2;BaseQRankSum=30.792;DB;DP=3756;Dels=0.00;FS=0.000;HaplotypeScore=158.0504;MLEAC=1;MLEAF=0.500;MQ=60.00;MQ0=0;MQRankSum=1.082;QD=11.10;ReadPosRankSum=3.174;SOR=0.553 GT:AD:DP:GQ:PL 0/1:2016,1733:3756:99:41736,0,48095

  • marcelcaraciolomarcelcaraciolo BrazilMember

    Additional information:

    I 've checked another region analysed and the strange happened:

    Even with nct deactivated I did not get the following SNP ( I did got with Unified Genotyper)

    chr12 48239835 rs1544410 C T 41707.77 . AC=1;AF=0.500;AN=2;BaseQRankSum=30.792;DB;DP=3756;Dels=0.00;FS=0.000;HaplotypeScore=158.0504;MLEAC=1;MLEAF=0.500;MQ=60.00;MQ0=0;MQRankSum=1.082;QD=11.10;ReadPosRankSum=3.174;SOR=0.553 GT:AD:DP:GQ:PL 0/1:2016,1733:3756:99:41736,0,48095

    image

    As you may see the screenshot the bam out at this region gave me any reads. What could be happening ?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Have a look in the Common Problems section of the documentation guide. There is an article about that problem which lists possible causes and solutions.

  • marcelcaraciolomarcelcaraciolo BrazilMember

    Hi Geraldine t hanks for your reply, exhausting all possibilites from the article, is there any chances that it could have any issues using the multi-threading option ?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    We have seen that happen but only with marginal calls, and then usually it is only semi-reproducible. Try running on just that region several times with different nct settings and see if you get the same result each time, to evaluate how reproducible it is.

Sign In or Register to comment.