How can I estimate the unfolded 2D SFS using dadi or any other tool?
46 views
Skip to first unread message
Tahir Ali
Feb 21, 2024, 9:55:02 PMFeb 21
to dadi-user
I hope you don't mind my asking what might seem like a simple question—I'm relatively new to estimating site frequency spectra based on derived alleles. I've been using a customized script and the est-sfs program (Keightley & Jackson, 2018) to estimate the unfolded SFS individually for four populations, incorporating three outgroups. However, I'm facing challenges in understanding how to estimate the unfolded 2D SFS for a pair of populations. Could someone offer some guidance? I'm particularly curious if there's a method in dadi that can help with this task Thank you for your support.
Ryan Gutenkunst
Feb 25, 2024, 4:30:42 PMFeb 25
to dadi...@googlegroups.com, Castellano, David - (dcastellano)
Hello Tahir,
My postdoc David Castellano (CC’d) explored using the est-sfs method for his analysis, so he may have more to say. But my understanding is that est-sfs is only built for single populations. If you dug into the code, you might be able to modify it to handle multiple populations; it’s conceptually straightforward.
But typically we include a parameter in our model fitting to estimate the rate of ancestral state misidentification, so we don’t worry too much about having perfect polarization. In dadi, users just specify a single estimated ancestral allele for each site, so there’s no probabilities like est-sfs. But I think the misid parameter does a good job of compensating for that.
Best,
Ryan
> On Feb 21, 2024, at 3:07 PM, Tahir Ali <touchs...@gmail.com> wrote:
>
> I hope you don't mind my asking what might seem like a simple question—I'm relatively new to estimating site frequency spectra based on derived alleles. I've been using a customized script and the est-sfs program (Keightley & Jackson, 2018) to estimate the unfolded SFS individually for four populations, incorporating three outgroups. However, I'm facing challenges in understanding how to estimate the unfolded 2D SFS for a pair of populations. Could someone offer some guidance? I'm particularly curious if there's a method in dadi that can help with this task Thank you for your support.
>
> --
> 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/d3fd6256-ee82-4d8e-8f2a-9b4b88f9de48n%40googlegroups.com.
My postdoc David Castellano (CC’d) explored using the est-sfs method for his analysis, so he may have more to say. But my understanding is that est-sfs is only built for single populations. If you dug into the code, you might be able to modify it to handle multiple populations; it’s conceptually straightforward.
But typically we include a parameter in our model fitting to estimate the rate of ancestral state misidentification, so we don’t worry too much about having perfect polarization. In dadi, users just specify a single estimated ancestral allele for each site, so there’s no probabilities like est-sfs. But I think the misid parameter does a good job of compensating for that.
Best,
Ryan
> On Feb 21, 2024, at 3:07 PM, Tahir Ali <touchs...@gmail.com> wrote:
>
> I hope you don't mind my asking what might seem like a simple question—I'm relatively new to estimating site frequency spectra based on derived alleles. I've been using a customized script and the est-sfs program (Keightley & Jackson, 2018) to estimate the unfolded SFS individually for four populations, incorporating three outgroups. However, I'm facing challenges in understanding how to estimate the unfolded 2D SFS for a pair of populations. Could someone offer some guidance? I'm particularly curious if there's a method in dadi that can help with this task Thank you for your support.
>
> 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/d3fd6256-ee82-4d8e-8f2a-9b4b88f9de48n%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages