Max mutation allele frequency against tumor purity

Question

[J.F.Jiang]

Hi all,

We are using GATK Mutect2 to call the somatic variants from tumor samples.

We found some of the variants had max allele frequency larger than 0.5, which made me quite confused.

Unless the sample is quite pure (100%), the theoretical allele frequency should be less than 0.5. While sometimes the frequency is 0.7.

This odds even appeared in ctDNA samples. As far as I know, most of the DNA are fragments originated from white cells, mixed with ctDNA from tumor cells. In this case, the frequency should be no more than 0.3. However, we can also find some variants with frequency larger than 0.3, which were confirmed somatic based on database (1000g, gnomAD, Clinvar, COSMIC, etc.).

Any comment will be appreciated.