Is it a bug for the Mutect2
I am using the Mutect2 (gatk-18.104.22.168) paired samples to call somatic mutation.
Here is my command line:
java -jar gatk-package-22.214.171.124-local.jar Mutect2 --reference hg19.fa -I tumor.sorted.rmdup.clipper.bam -tumor test1 -I normal.sorted.rmdup.clipper.bam -normal test2 --germline-resource gnomad.exomes.2.vcf.gz --panel-of-normals pon.vcf.gz -L test.DNA.bed -O test.match.vcf.gz
Here is the result generated by Mutect2:
chr2 25457242 . C T . . DP=967;ECNT=1;NLOD=114.19;N_ART_LOD=-9.232e-01;POP_AF=2.197e-04;P_GERMLINE=-1.116e+02;TLOD=267.27 GT:AD:AF:F1R2:F2R1:MBQ:MFRL:MMQ:MPOS:SA_MAP_AF:SA_POST_PROB 0/1:406,94:0.204:199,39:207,55:38:224,229:60:15:0.182,0.172,0.188:4.684e-03,0.012,0.983 0/0:397,8:0.046:195,5:202,3:34:226,248:60:1
The AF of tumor sample is 0.204 that looks at little overestimated, bit the normal sample is 0.046 which is far overestimated compared with IGV (AF=1/(404+1), only 1 read carry this mutation. But Mutect2 result shows there are 8 reads carry this mutation).
Does mutect2 do re-alignment before calling any mutation? I cannot figure out where those extra mutant reads come from.