Stretching Data Suitability

[David Dayan]

I have a general/exploratory question about the suitability of some SNP data to dadi.

I've begun working with a small SNP panel (~500 SNPs) in a new conservation-oriented position. Most of these are neutral, unlinked SNPs distributed throughout the entire genome, but a subset of ~30 are from a genomic region of intense interest for conservation and management given it's strong association with an important phenotype and evidence that it is under selection. The average intermarker distance in this region is about 5kb, LD patterns are complex and vary across population and ecotypes.

I'd like to infer the demographic history of two sympatric ecotypes that diverge at this genomic region using the data I have in hand. In short, my goal is to understand if a demographic model with two epochs of gene flow is a better fit to the data than a model with single epoch. I realize it's kind of a big question, but I'm curious how the very uneven distribution of information along the genome (and strongly biased SFS) will influence inference of demography in dadi. Is it worth taking an SFS approach, or should I look to LD or something else?

Thanks and I hope this generates some fun discussion,

David

[Suany Quesada]

Hello.

Were you able to solve your question? I have a similar situation. I have a problem with running dadi with SNP panel data:

I am running 2D and 3D models with chip sequencing SNPs (SNP panel) of a mammal, however I get a pattern of sfs with empty spaces. The data are filtered in term of LD. Is the obtained pattern normal? Is it due to the sequencing technique? Any suggestions?

I attach two examples (Evaluation of founder effect without migration in a 2D model and evaluation of divergence with migration).

Thank you very much in advance for your suggestions and comments!