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!

Can the new Mutect2 (version > support a GVCF output?


I was wondering if the latest version of Mutect2 can now support a GVCF output.

According to a blog titled "Short Variant Caller Roundup" from last February:
"The GATK release introduces a beta version of GVCF mode for Mutect2 designed for mitochondrial variant calling...".
Is that true for chromosomal variants too?

I ran the following command with the gatk- version:
'java -jar $GATKMutect2 -R $ref -I $outdir$outname"_split.bam" --germline-resource $afgnomad --panel-of-normals $pon --af-of-alleles-not-in-resource 0.0000625 -tumor $outname --emit-ref-confidence GVCF -O $outdir$outname".g.vcf" 2>$outdir$outname.Mutect2.log'

and the output has a GVCF format, but I'm not sure if this is a trusted output, given that the documentation for Mutect2 does not specify that it's capable of producing GVCF files.

'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT a142506d-9162-4978-88ff-6ab600a058ab
chr20 1 . N <NON_REF> . . END=87644 GT:DP:MIN_DP:TLOD 0/0:0:0:0.00
chr20 87645 . T <NON_REF> . . END=87734 GT:DP:MIN_DP:TLOD 0/0:2:1:-4.765e-01
chr20 87735 . T <NON_REF> . . END=96051 GT:DP:MIN_DP:TLOD 0/0:0:0:0.00
chr20 96052 . A <NON_REF> . . END=96193 GT:DP:MIN_DP:TLOD 0/0:1:1:-4.765e-01
chr20 96194 . G <NON_REF> . . END=113212 GT:DP:MIN_DP:TLOD 0/0:0:0:0.00
chr20 113213 . T <NON_REF> . . END=113287 GT:DP:MIN_DP:TLOD 0/0:1:1:-3.010e-01
chr20 113288 . G <NON_REF> . . END=113307 GT:DP:MIN_DP:TLOD 0/0:0:0:0.00
chr20 113308 . C <NON_REF> . . END=113366 GT:DP:MIN_DP:TLOD 0/0:1:1:-3.010e-01
chr20 113367 . C A,<NON_REF> . . DP=1;ECNT=1;MBQ=0,40,0;MFRL=0,170,0;MMQ=60,60,60;MPOS=15,50;POPAF=4.20,4.20;TLOD=4.10,-1.761e-01 GT:AD:AF:DP:F1R2:F2R1:SB 0/1/2:0,1,0:0.500,0.250:1:0,0,0:0,1,0:0,0,0,1
chr20 113368 . A <NON_REF> . . END=113382 GT:DP:MIN_DP:TLOD 0/0:1:1:-3.010e-01
chr20 113383 . T <NON_REF> . . END=131445 GT:DP:MIN_DP:TLOD 0/0:0:0:0.00
chr20 131446 . G <NON_REF> . . END=131500 GT:DP:MIN_DP:TLOD 0/0:1:1:-3.010e-01
chr20 131501 . A <NON_REF> . . END=131536 GT:DP:MIN_DP:TLOD 0/0:0:0:0.00
chr20 131537 . A <NON_REF> . . END=131603 GT:DP:MIN_DP:TLOD 0/0:1:1:-3.010e-01'


  • bshifawbshifaw Member, Broadie, Moderator admin

    Hi @Christos,

    The tool docs still shows the --emit-ref-confidence parameter in BETA for the latest version of the gatk (

  • ChristosChristos Member
    They show it for older versions too, even though they definitely don't support GVCF. In another thread in this forum one of the admins said that this is just because Mutect2 and HaplotypeCaller share the same argument list on the docs.
  • bshifawbshifaw Member, Broadie, Moderator admin

    Just confirmed with the dev team, that's right, because Muctect2 uses haplotypecaller to perform calling the arguments are shared between the tools. The beta indicator for --emit-ref-confidence in the tools docs is to indicate the beta status for running Mutect2 -GVCF on mitochondrial DNA. Running this argument in haplotypecaller for its intended purpose is fine.


    GVCF mode in M2 should only be used for mitochondrial calling. Multiple tumors in M2 should be called with -I tumor1.bam -I tumor2.bam -I tumor3.bam -I normal.bam

Sign In or Register to comment.