Max diversity (instead of mean) for evaluating IBD in spatial simulations

[zach.h...@tamu.edu]

Hi all!

I'm trying to understand how sampling density and % of range sampled in spatial simulations impacts our ability to infer isolation-by-distance (IBD) curves. Because of this, we're interested in the max genetic diversity of the population (instead of the mean), since this better captures (I think) the differences between most geographically distant individuals. 

For sample sets, this is easy, just make a pairwise matrix. But for the entire population, this becomes a very large matrix. Perhaps there is no easier way, but I thought I'd pose the question here: does anyone know a way to grab the max genetic diversity from the total population without needing to compute the full matrix? Something like identifying the most distantly related individuals in the tree? 

Thanks for any help you can provide! 

Zach

[Peter Ralph]

Hi, Zach - good question! We don't have an easy way to do this
specific thing, although it's an interesting suggestion. Two other
suggestions:
- just look at divergence between furthest-geographically individuals
- look at `segregating_sites` instead, as that measures "total
diversity" (and, total area of the tree), albeit in a different way

Also, we're trying to move discussion over to github discussions;
perhaps could you post this at
https://github.com/tskit-dev/tskit/discussions if you'd like to
discuss more?

--peter

> --
> You received this message because you are subscribed to the Google Groups "msprime-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to msprime-user...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/msprime-users/8a5afac4-88d0-44b0-9425-ace2581090ffn%40googlegroups.com.