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!

Why variant in germline-resource not filtered.

Hi,
I have paired samples and I called them separately using following command, got a variant call in normal sample, not in tumor sample.

chr8:139165272 should be pre-filtered because it is in the germline-resource database. But how it was not filtered in the Mutect2 result?

normal sample:

chr8    139165272       .       TG      CA      .       .       DP=115;ECNT=2;MBQ=36,40;MFRL=218,208;MMQ=60,60;MPOS=43;POPAF=7.30;SAAF=0.323,0.333,0.354;SAPP=0.026,0.011,0.963;TLOD=147.15    GT:AD:AF:DP:F1R2:F2R1   0/1:73,40:0.356:113:30,21:43,19

command line:

~/gatktools/gatk-4.1.2.0/gatk Mutect2 \
-R ~/hg19/gatk_bundle/ucsc.hg19.fasta \
-I tumor.recalibrated.bam -tumor tumor \
-L chr8:139165272 --interval-padding 250 \
--germline-resource ~/af-only-gnomad.raw.sites.hg19.vcf.gz \
--tmp-dir tmpdir -O tmp.vcf

Thank you.

Tagged:

Answers

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin
    edited September 2019

    Hi @xiucz

    If I understand correctly, you ran Mutect2 on the "normal" sample in the tumor only mode with a germline-resource and Mutect2 did not filter that variant? Is that correct?

    Also please post the gnomad vcf record of this variant.

  • xiuczxiucz Member

    @bhanuGandham ,

    Yes, I think it should be filtered both in the tumor and normal, please note that the germline-resource is the raw one, not the small one.

    zcat ~/af-only-gnomad.raw.sites.hg19.vcf.gz | grep -w 139165272
    chr8    139165272   rs7835712   T   C   16355619.74 PASS    AC=28043;AF=0.69
    

    Thank you.

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin
    edited September 2019

    Hi @xiucz

    I am a little confused. Would you please clarify my questions:
    1) What is the purpose of running Mutect2 on Normal samples in tumor only mode with a germline-resource? Also you shared the commandline for Mutect2 on tumor sample and not the normal sample.
    2)

    got a variant call in normal sample, not in tumor sample.

    What do you mean by this?
    3) Did you perform any filtration on your data? Take a look at this doc: https://software.broadinstitute.org/gatk/documentation/article?id=24057

    Post edited by bhanuGandham on
  • xiuczxiucz Member

    Hi, @bhanuGandham
    1) Purpose: Sometimes, we don't have a matched normal sample.So I did some test in tumor-only mode, but apparently there were lots of false positive variants in tumor-only mode, thus I added the germline-resource parameter to get germline variants pre-filtered. Then, I used the following commandline:

    tumor sample:
    ~/gatktools/gatk-4.1.2.0/gatk Mutect2 \
    -R ~/hg19/gatk_bundle/ucsc.hg19.fasta \
    -I tumor.recalibrated.bam -tumor tumor \
    -L chr8:139165272 --interval-padding 250 \
    --germline-resource ~/af-only-gnomad.raw.sites.hg19.vcf.gz \
    --tmp-dir tmpdir -O tumor.tmp.vcf
    
    normal sample:
    ~/gatktools/gatk-4.1.2.0/gatk Mutect2 \
    -R ~/hg19/gatk_bundle/ucsc.hg19.fasta \
    -I normal.recalibrated.bam -tumor normal \
    -L chr8:139165272 --interval-padding 250 \
    --germline-resource ~/af-only-gnomad.raw.sites.hg19.vcf.gz \
    --tmp-dir tmpdir -O normal.tmp.vcf
    

    2)
    As a consequence,

    $ grep -w 139165272 normal.tmp.vcf
    $ chr8    139165272       .       TG      CA      .       .       DP=115;ECNT=2;MBQ=36,40;MFRL=218,208;MMQ=60,60;MPOS=43;POPAF=7.30;SAAF=0.323,0.333,0.354;SAPP=0.026,0.011,0.963;TLOD=147.15    GT:AD:AF:DP:F1R2:F2R1   0/1:73,40:0.356:113:30,21:43,19
    
    $ grep -w 139165272 tumor.tmp.vcf
    $
    (i got nothing)
    

    3) I did not perform any filtration nextstep yet, because I wanted to know why I failed in calling the site out in my tumor sample first. Should I use GGA mode to call the site ?

    I feel sorry for disturbing you so much and thank you very much.

    Xiucz.

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin

    HI @xiucz

    Sorry I wasn't clear, what is the purpose of running tumor-only mode on Normal Samples?

  • xiuczxiucz Member

    Hi @bhanuGandham ,
    I will put myself in a simple way, let us take my purpose aside.
    Using the same commandline, why the site cannot be pre-filtered in different samples, although this site is recored in germline-resource.

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin

    Hi @xiucz

    Sorry I got distracted by what it is that you were trying to achieve. But I see your point. I will look into it and get back to you.

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin

    HI @xiucz

    So we suspect that this might be happening because the variant record you showed us in the normal sample, chr8 139165272 . TG CA is a MNP while gnomad germline resource has only SNPs. This might be why it does not get filtered.
    Try to set max-mnp-distance to 0 and run mutect2 again. Let us know if that resolves the issue.

  • xiuczxiucz Member

    [email protected]

    after setting max-mnp-distance to 0, I still got this site:

    ~/gatktools/gatk-4.1.3.0/gatk Mutect2 \
    -R ~/hg19/gatk_bundle/ucsc.hg19.fasta \
    -I normal.recalibrated.bam -tumor normal \
    -L chr8:139165272 --interval-padding 250 \
    --germline-resource ~/af-only-gnomad.raw.sites.hg19.vcf.gz \
    --tmp-dir tmpdir -O normal.tmp2.vcf --max-mnp-distance 0
    

    the MNV has been split into two SNVs,

    chr8    139165272   .   T   C   .   .   DP=169;ECNT=3;MBQ=20,20;MFRL=213,201;MMQ=60,60;MPOS=45;POPAF=0.161;TLOD=221.95  GT:AD:AF:DP:F1R2:F2R1:PGT:PID:PS:SB 0|1:106,61:0.367:167:43,33:60,28:0|1:139165272_T_C:139165272:48,58,28,33
    chr8    139165273   .   G   A   .   .   DP=169;ECNT=3;MBQ=20,20;MFRL=213,201;MMQ=60,60;MPOS=44;POPAF=0.451;TLOD=221.95  GT:AD:AF:DP:F1R2:F2R1:PGT:PID:PS:SB 0|1:106,61:0.367:167:40,33:63,27:0|1:139165272_T_C:139165272:48,58,28,33
    

    This gnomad germline resource site seems to not be prefiltered still.

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin

    Hi @xiucz

    That is weird. After splitting into two SNVs the chr8 139165272 . T C variant should have been filtered. I will check with the dev team and get back to you.

  • 29043594952904359495 Member ✭✭

    I have another question, if we use --max-mnp-distance 0 in the Mutect2 command, will all variants are snv, and no mnp, there is of course not right I guess.
    @xiucz ,have you checked this, thanks a lot

  • xiuczxiucz Member

    @2904359495 yes, there will be no MNVs.

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin

    Hi @xiucz

    Have your run FilterMutectCalls on the results of Mutect2? If not, then you should do that. That will resolve this issue.

  • xiuczxiucz Member
    edited November 2019

    :D@bhanuGandham,

    following your insiting advice,I got PASS result still....

    Today, I got another pair samples' result:

    Using MuTect:

    #normal sample tumoronly-mode's mutect result:
    CAGxACG T       C       normal none    0       DBSNP+COSMIC  COVERED  1       1       0       0       0       202     0       -41.248107      10.104493       7.209222        10.383325       0.020101      0.02     10.078254       199     195     4       6319    167     60      60      0       0       TT      -0.693147       0       0       0     00       0       0.777877        0.575503        (95,166,3,4)    56.5    24.5    92.5    24.5    0       possible_contamination  REJECT
    
    #tumor sample tumoronly-mode's mutect result:
    CAGxACG T       C       tumor none    0       DBSNP+COSMIC  COVERED  1       1       0       0       0       146     0       257.594437      257.765154      118.715727      185.997276      0.462069      0.02     4.895241        144     78      67      2468    2692    60      60      0       0       TT      -0.693147       0       0       0     00       0       1       1       (32,64,32,48)   98      23      51      23      0               KEEP
    
    #tumor-normal-mode's mutect result:
    chr9    133738357       rs121913461     T       C       .       REJECT  DB      GT:AD:BQ:DP:FA  0:198,4:.:203:0.020     0/1:78,67:40:145:0.462
    

    Using GATK4.1.3.0 Mutect2:

    #tumor sample tumoronly-mode's result:
    chr9    133738357       rs121913461     T       C       .       PASS  AC=2;AF=0.400;AN=5;CONTQ=93;DB;DP=146;ECNT=1;GERMQ=12;MBQ=32,41;MFRL=234,244;MMQ=60,60;MPOS=45;POPAF=7.30;SAAF=0.424,0.444,0.461;SAPP=0.097,7.159e-03,0.896;SOMATIC;TLOD=243.06;VT=SNP;set=Intersection GT:AD:AF:BQ:DP:F1R2:F2R1:FA:OBAM:OBAMRC:OBQ:OBQRC:SS    0/1:76,65:0.464:.:141:35,31:41,34:.:false:false:60.26:100.00   0/1:78,67:.:40:145:.,.:.,.:0.462:.:.:.:.:2      0:0,0:.:.:0:.,.:.,.:0.00:.:.:.:.:0
    
    #no result in normal
    
    #tumor-normal-mode's result:
    chr9    133738357       .       T       C       .       PASS    CONTQ=93;DP=350;ECNT=1;GERMQ=417;MBQ=32,41;MFRL=238,243;MMQ=60,60;MPOS=45;NALOD=-7.776e+00;NLOD=41.82;POPAF=6.00;SAAF=0.424,0.444,0.461;SAPP=0.097,7.159e-03,0.896;TLOD=243.06        GT:AD:AF:DP:F1R2:F2R1:OBAM:OBAMRC:OBF:OBP:OBQ:OBQRC     0/0:195,4:0.025:199:91,2:104,2:false:false      0/1:76,65:0.464:141:35,31:41,34:false:false:.:.:60.26:100.00
    

    It seemed that in tumor-only mode, only the site have a low VAF that can be prefiltered by the germline resource. If the site have a high VAF, it cannot be prefiltered even in the germline-resource database.

    So, is this a bug i, or a threshold in the prefilter step?

  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin
    edited November 2019

    Hi @xiucz

    Can you please post the exact FilterMutectCalls command you used and please highlight the examples from above that are from the output of FilterMutectCalls.
    Mutect2 does some pre-filter but that is not sufficient and hence we recommend always evaluating the variant calls after executing FilterMutectCalls tool.

Sign In or Register to comment.