Interpreting residuals

[Miles Anderson]

Hi, Ryan

I am running a small test on two populations from a larger dataset using a split_mig model and wanted to ask how to interpret the residuals calculated just to be sure. It seems that the glaring issue is that singletons in both populations are overestimated and shared singletons are underestimated. Is this interpreting the graph correctly? Do the residuals otherwise seem reasonable?

The split_mig model is almost certainly oversimplified as we have done some environmental niche modeling that suggests these populations would have undergone huge range contraction in the last 10,000 years followed by massive proliferation and probably increased gene flow after that time. Would such a population contraction explain the apparent overprediction of rare alleles in the individual populations, and then a growth event followed by increased gene flow explain the underprediction of shared rare alleles?

Or would you expect something else is wrong? The confidence intervals in any case have a negative lower bound for migration, so a custom model will likely need to account for a different migration scenario.

Best,

Miles

[Ryan Gutenkunst]

Hello Miles,

Your interpretation of the residual plot is correct. I agree the other portions of the plot look pretty good.

A population size contraction would indeed be consistent with the model over-predicting rare alleles. The interpretation of the under-predicted shared alleles is a little more complex, but your logic is sound.

Best,
Ryan

> --
> You received this message because you are subscribed to the Google Groups "dadi-user" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to dadi-user+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/dadi-user/d6dd504a-421d-4249-a787-4b03c80ffffan%40googlegroups.com.
> <CV1xCV2_demo_plot.png>