How rare are somatic mutations?

Question

[Kermit]

In Foundation Medicine's Tumor Mutation Burden (TMB) study... https://genomemedicine.biomedcentral.com/articles/10.1186/s13073-017-0424-2

They claim a median of 3.6 somatic mutations per Mb (million) in metasticised samples. Despite the fact that the experiment filtered out germline and dbSNP variants, I was expecting a much higher mutation rate due to the fact that these individuals are highly likely to have loss of function in mismatch repair and replication repair genes.

In germline, I believe the rule of thumb is something like 1 in 1,000 bp is a variant. Meaning 1,000 mutations per Mb.

What is the rate of mutation in somatic variant calling?

[steve]

depends completely on your variant calling methods and your variant filtering methods. You could call hundreds or thousands of variants on a sample but then filter down to single digit final variant list to use for TMB. Or anything in between.

[ATpoint]

Also depends on the disease. There are high-mutation cancers like lung or skin cancer and low-mutation ones like Acute Myeloid Leukemia. There is no rule of thumb that would acurately reflext the complexity of both the variety of diseases and variant calling methods. in DLBCL (B cell lymphoma) I've seen patients with somatic mutations ranging from like 1000 to 20000. Also depends on filtering. Some somatic variants might turn out to be common polymosphisms but depending on the genome build you use these might or might not be annotated in the respective databases you filter against, even rhough the caller might have flagged them as somatic. That having said, reference genome of course matters as well.