Recording mutation frequencies through time

[leanngward]

I am working with simplified trees generated by SLiM. I have also overlaid neutral mutations using msprime.SLiMMutationModel.

I have been able to use the Python API to find the frequencies of mutations at the final generation (i.e., the time at which the tree was recorded):

sites = ts.sites()
for site in sites:
       tree = ts.at(site.position)
       for mutation in site.mutations:
               num_samples = tree.num_samples(mutation.node)
               total_samples = tree.num_samples()
               frequency = num_samples / total_samples

Question: Is there a way to find mutation frequencies at every generation in tree?

From what I could find in similar posts, I wonder if I might need to avoid simplifying trees in SLiM. However, I also noticed that the above code no longer returns correct frequencies when trees are not simplified, so I will have to make some adjustments. (Also, to clarify, "correct" frequencies are validated from mutation frequencies tracked during the SLiM simulation for non-neutral mutations.)

Thanks!

[petrelharp]

Question: Is there a way to find mutation frequencies at every generation in tree?

Depends what you mean here. If you mean "Is there a way to find mutation frequencies at every generation?", the answer is "no", because you've simplified, as you suspect. The way I'd do this is not by avoiding simplifying, but by Remembering all individuals alive at the times you want the frequencies at.