I was trying mutect2 to call somatic variant and found out that, by default, it uses 6.3 as tumor log odds (LOD) filtering cutoff. The formula is:
"Assuming somatic mutations to occur at a rate of ~1 per Mb. Therefore, LOD_T > log_{10} (0.5 \times 10^{-6} ) \approx 6.3 $$, which guarantees that our false positive rate, due to noise in the tumor, is less than half of the somatic mutation rate"
Here is the GATK link https://gatkforums.broadinstitute.org/gatk/discussion/4463/how-mutect-identifies-candidate-somatic-mutations
It seems to me the tumor mutation frequency is too arbitrary. The mutation frequency varies between different cancer type by two magnitude.
Does this make sense? Or I misunderstood it?
You usually want to have a fixed assay-specific error rate. They just used this example to put the number in context. But sure, if you can tolerate more false positives when you are working in CRC, nothing prevents you from setting cutoffs accordingly.
Log in to answer this question.
