All Models Collapsing at same place

[Reuben Ockman]

Hello Dadi Users, 

I'm new to demographic modeling and so I have tried to read some of the conversations in this forumn for some help, but wanted to reach out and ask for advice. 

I ran a 1D model simulation from Linked SNP data. I was able to make the SFS and was modelling inbreeding so went with the subsampling approach. While I have tested 4 models so far, all of them (except for the neutral model) continue to collapse at the same place and so produce the same residual plots and I'm not sure how to intepret them or to improve them. I know that when dealing with Linked SNPs you cannot simply use the composite likelyhoods in order to compare the models. 

 I guess I'm wondering if this is normal and also if you would suggest LD Prunning before running these models 

Any advice would be greatly appreciated!

Best,

Reuben

[Ryan Gutenkunst]

Hello Reuben,

I’m not sure what you mean by “collapse at the same place”. The drop at frequency 40 is due to the folding of the frequency spectrum, because your data aren’t polarized. In general, the exponential growth and instantaneous two-epoch models are not typically distinguishable. There will be parameter sets that give almost the exact same fit. Qualitatively, there’s not a strong signal of inbreeding in your data.

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 visit https://groups.google.com/d/msgid/dadi-user/c76c238d-350d-4c1d-8685-93cf9bb73e04n%40googlegroups.com.
<Aegean_1D_snm_comparison.png><Aegean_1D_two_epoch_demo_plot.png><Aegean_1D_two_epoch_inbreeding_demo_plot.png><Aegean_1D_growth_demo_plot.png>