Huge difference in sample size between population

[durandka...@gmail.com]

Hi,

I have Two populations , pop1 n=13 and pop2 n=78 .

My pop are diploid so i did the fs with projections =[26,160]

But my SFS is quite empty.

Should it be better to have population of the same sample size ? and could i use projections =[26,26] to have a better SFS ?

Best

Karine

[Ryan Gutenkunst]

Hello Karine,

Projecting downward will reduce your power, so if you won’t have missing data issues, I would not do it. Your spectra does look very strange. It is unbelievable to me that pop2 has no stations with frequency greater than 5/160. There may be some issues in your data processing to reach the sfs stage.

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/1909e80c-18fc-45d0-a7d1-b96e1e605b1an%40googlegroups.com.
> <Figure_1.png><Figure_2.png>

[durandka...@gmail.com]

Hi

Thanks, I use the script convert_vcf_to_dadi_input.pl for my vcf, and SFS looks better!

Best

[Ryan Gutenkunst]

Hello Karine,

This is definitely an improvement. It’s not clear to me why you have so many missing low-frequency alleles in your data those. That is likely highly biasing your results. If this is unavoidable, you’ll need to mask those entries in the spectrum.

Best,
Ryan

> To view this discussion on the web visit https://groups.google.com/d/msgid/dadi-user/9d4279b3-8b08-4261-8d2d-f968821ec006n%40googlegroups.com.
> <SI_2021_12_6_122528_SI.pdf>

[durandka...@gmail.com]

Thanks for your help!