Expected mutation probability calculation?

[ugh-astya]

I have gone through the bioRxiv manuscript but find it vague and confusing. I am trying to reporoduce the results but am running into problems.
How is expected mutation probability calculated?

Using downsampled genomes we find the observed probability of a base mutating to another in a particular context.
Observed probability= No. of AAA->ATA mutations / (Occurence of AAA in the whole genome * 3)
Next, we convert it to mutability by multiplying obs. probability with scaling factor S.
S=Total no. of bases * Mutation rate (1.2e-8) / Total no. of mutations
Here, is the total no. of mutations calculated from the downsampled 1000 genomes or the complete 75000 genomes?

When fitting the mutability calculated above to proportion (exponential capped at 1), how is the proportion calculated?
Is it No. of AAA->ATA mutation / Total no. of AAA mutated ? If not then what is it? The manuscript uses proportion interchangeably referring to different proportions, so it gets very confusing
To my understanding the above is calculated using 75000 genomes.