What I understand is Tumor mutation burden (TMB) is number of unique mutation in a sample/ the size of the capture

and variant allele fraction (VAF) is tumour allele count / coverage and can be infer as the size of the mutant clone. correct?

Let say I am sequencing normal tissue to compare the somatic mutation load between 2 samples and the maf file is as follow for sample A and B with capture size of 10Mb:

ID refcount altcount    coverage    TumorVAF
A  1268        1088         2356    0.46179966
A  1260        1237         2497    0.495394473
B  2113        20           2133    0.009376465
B  793         625          1418    0.440761636

Tumor mutation burden for both A and B would be 2 mutations/10Mb = 0.4/Mb

But if I want to find the mutation load or fraction of mutant cells in the normal tissue, can it be calculated as
sum of tumour allele count/ sum of coverage

i.e. 1088+1237/2356+2497= 0.47 for sample A
20+625+2133+1418= 0.181 for sample B?

Or it can't because we cannot assume all cells and regions were sequenced evenly?

I know TMB is usually reported but wouldn't TMB under-estimate or over-estimate in this case the mutation burden in sample B? when actually only a small fraction of cells are mutated?