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!

How to explain a variant present in PON is classified as PASS in the output using MuTect2

How to explain a variant present in PON is classified as PASS in the output (do not filtered as panel_of_nomal) using Mutect2.

The VCF corresponding to this variant in the PON file (called normal.vcf) is

15 9100143 . TG T . PASS AC=16;AF=0.500;AN=32;ECNT=1;MAX_ED=.;MIN_ED=.;NLOD=0.00;RPA=4,3;RU=G;STR;set=Intersection GT:AD:AF:ALT_F1R2:ALT_F2R1:QSS:REF_F1R2:REF_F2R1 0/1:0,65:1.00:34:31:0,2037:0:0 0/1:0,84:1.00:43:41:0,2754:0:0 0/1:0,66:1.00:28:38:0,2082:0:0 0/1:1,46:0.976:17:29:29,1394:0:1 0/1:0,78:1.00:28:50:0,2414:0:0 0/1:0,62:1.00:24:38:0,1862:0:0 0/1:0,74:1.00:29:45:0,2261:0:0 0/1:0,58:1.00:31:27:0,1789:0:0 0/1:0,73:1.00:23:50:0,2219:0:0 0/1:0,56:1.00:20:36:0,1680:0:0 0/1:0,51:1.00:21:30:0,1584:0:0 0/1:0,71:1.00:41:30:0,2197:0:0 0/1:0,66:1.00:37:29:0,2001:0:0 0/1:0,88:1.00:40:48:0,2635:0:0 0/1:0,61:1.00:23:38:0,1898:0:0 0/1:0,40:1.00:15:25:0,1195:0:0

MuTect2 is called as

java -Djava.io.tmpdir=_tmp/javaio -Xmx80G -jar GenomeAnalysisTK.jar -T MuTect2 -nct 20 \
-I:tumor _in/EU1.bam -PON _in/normal.vcf -o _out/EU1.vcf \
-R ../../../REFERENCES/GATKResourceBundle/GRCm38.fa \
-L ../../../REFERENCES/enrichment/Regions.interval_list --dbsnp ../../../REFERENCES/GATKResourceBundle/Mus_musculus.vcf.gz

And the output line is

15 9100143 . TG T . PASS ECNT=1;HCNT=1;MAX_ED=.;MIN_ED=.;NLOD=0.00;RPA=4,3;RU=G;STR;TLOD=182.36 GT:AD:AF:ALT_F1R2:ALT_F2R1:FOXOG:QSS:REF_F1R2:REF_F2R1 0/1:0,60:1.00:29:31:.:0,1895:0:0

Tha same happened in other 14 tumor samples. All of then have called this variant as somatic instead of germline.

Any idea?

Note that the normal.vcf (PON) file was created not just calling normal tissues un --artifact_mode); therefore has genotype field of many samples. ¿Could be this the problem?.

Thank in advance

Tagged:

Best Answer

  • SheilaSheila Broad Institute ✭✭✭✭✭
    edited October 2016 Accepted Answer

    @jorgebarrera
    Hi,

    MuTect2 takes into account the PON and dbSNP variants when making a somatic variant call, however, it also takes into account how much evidence there is for the variant in the tumor samples. If there is a lot more evidence for the variant in the tumor samples compared to dbSNP and the PON, the variant will be called with high confidence. Have a look at this thread for some extra information.

    -Sheila

Answers

  • SheilaSheila Broad InstituteMember, Broadie ✭✭✭✭✭
    edited October 2016 Accepted Answer

    @jorgebarrera
    Hi,

    MuTect2 takes into account the PON and dbSNP variants when making a somatic variant call, however, it also takes into account how much evidence there is for the variant in the tumor samples. If there is a lot more evidence for the variant in the tumor samples compared to dbSNP and the PON, the variant will be called with high confidence. Have a look at this thread for some extra information.

    -Sheila

  • jorgebarrerajorgebarrera SpainMember

    Probably I misunderstanding something but I thought that PON intended to filter germline variants and technical artifact out. Therefore, If a variant is in PON because it is a germline variant then the evidence of such variant in tumor samples is irrelevant. It is germline variant. Am I right?

  • SheilaSheila Broad InstituteMember, Broadie ✭✭✭✭✭

    @jorgebarrera
    Hi,

    Yes, the PON is there to help filter out germline variants. But, it is possible for somatic mutations to appear in the PON due to contamination. Or, some "non-tumor" samples may actually contain tumors. For those reasons, we take into account the fact that the variant is seen in the tumor. When you submit a PON, the evidence in the tumor sample must be more than the normal threshold when you do not submit a PON.

    I hope that helps.

    -Sheila

  • SheilaSheila Broad InstituteMember, Broadie ✭✭✭✭✭

    @jorgebarrera
    Hi again,

    Would you be able to submit a bug report so I can test locally. After talking to the team, I think this may be something we can improve on. Instructions are here.

    Thanks,
    Sheila

Sign In or Register to comment.