Using f3 to test gene flow from ghost species

[kiwoong-nam]

I ask you this question to check if my logic makes sense.

Suppose this phylogeny (A, (B,C)), and we want to test gene flow from ghost species outside of this tree to C. In this case, we expect the overrepresentation of this genotype: 0, 0, 1 for A, B, and C, respectively. Then, f3 = (c-a)(c-b)=(1-0)(1-0) = 1 > 0.

Thus, if f3 > 0, it supports the existence of gene flow from ghost species.

Does it makes sense?

[bumblenick]

This idea doesn't work because f_3 > 0 wit hout any adnixture at all

Nick

[kiwoong-nam]

if I understand your answer correctly, the admixture will surely cause f3 > 0. But, f3 > 0 does not necessarily mean admixture.

Simple counting of genotype combinations would probably work better..

Thank you for your answer.

[alanrogers]

That admixture will inflate the frequency of 110 as well as 001. You might get more power by using this additional piece of information. Alan Rogers

…

On Fri, Feb 9, 2024 at 3:10 AM kiwoong-nam ***@***.***> wrote: I ask you this question to check if my logic makes sense. Suppose this phylogeny (A, (B,C)), and we want to test gene flow from ghost species outside of this tree to C. In this case, we expect the overrepresentation of this genotype: 0, 0, 1 for A, B, and C, respectively. Then, f3 = (c-a)(c-b)=(1-0)(1-0) = 1 > 0. Thus, if f3 > 0, it supports the existence of gene flow from ghost species. Does it makes sense? — Reply to this email directly, view it on GitHub <#103>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABRR6SQIZEPTO7ZCQLEKU33YSXYYJAVCNFSM6AAAAABDBGMIOWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGEZDMOBSHA4DKNY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>