What is population branch statistic (PBS)?

Question

[beausoleilmo]

In an article about Tibetan adaptation, they talk about a population branch statistic. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3711608/

By comparing the three pairwise FST values between these three samples, we can estimate the frequency change that occurred in the Tibetan population since its divergence from the Han population (5, 9). We found that this population branch statistic (PBS) has strong power to detect recent natural selection (Fig. S3).

I was wondering how it was calculated and what that means (lower vs larger values).

It seems that it can tell when there was natural selection at a certain time.

What are the assumptions of this model?

Answer 1

[GabrielMontenegro]

If you'll go to the supplementary material you will find plenty of information regarding the statistic. It is a simple measure of distance between three populations (branch length T from Cavalli-Sforza, 1969) using FST. The PBS basically measures if there is an allele with extreme frequency compared to two other populations. It is VERY similar to Shriver et al. 2004.

[beausoleilmo]

Cool! They actually give a good description. Do you know why they say that it's more powerful do detect recent selection sweeps?

It should have power, for example to detect incomplete selective sweeps, a type of selection that is highly relevant here and which most other statistics based on the SFS have little power to detect.

[GabrielMontenegro]

If I remember correctly, they did simulations to estimate the power. I think they compared to Tajima's D or other commonly used selection statistics. You could check that section. But usually, FST-based methods will have good power to detect recent selection, because you can use a closely related population to estimate FST. Even if the split was recent and adaptation occurred quite recently i.e. genome-wide FST will be small, the FST at the selected loci will be an outlier.