Restricted simulation that can be simulated backwards and forwards

[hyanwong]

At PopGroup57, I chatted for a while to Moritz Otto from Univ of Cologne (Köln), who recently published this paper: https://doi.org/10.1016/j.tpb.2023.08.001

This describes a method for using msprime to simulate tandem duplications in a recombining population, by treating each class of counts (1, 2, 3, 4, etc duplicated "genes") as a separate subpopulation (and not allowing mutations within each "gene")

I think we could create both a forward and a simulation that does this and saves the result as a GIG. Then we can test the results from both sorts of simulation against the expectations from Moritz's msprime simulations. He's happy for us to do this, and might be interested in the GIG concept we've started to build.

[duncanMR]

That is very helpful! When I did the literature review, I struggled to find examples of coalescent simulation tools that could simulate any structural variation except for inversions. Given all the technical challenges of simulating duplications, it will be great to have something to check our results against.

[hyanwong]

Moritz is on GitHub at @Moritz-Otto, and the simulation stuff he has put together is at https://github.com/Moritz-Otto/motto-structured_coalescent