OBJECTIVES Wright defined FST as a measure of genetic differentiation. Cockerham developed an estimator of FST based on binary indicators in an ANOVA framework. Here, we address 2 issues regarding the estimation of FST. First, we derive a new estimator of FST based on the ANOVA framework using the doubly truncated normal distribution as an approximation of the binomial distribution to estimate variances. Second, we consider the impact of de novo mutation on FST estimation. METHODS We compare our estimator to Weir and Cockerham's estimator via computer simulation. We apply our estimator to whole genome sequence data from the 1000 Genomes Project. We use chimpanzee whole genome sequence data to ascertain for ancestral polymorphisms. RESULTS By simulation, our new estimator is less biased than Weir and Cockerham's estimator for comparison of two subpopulations and is systematically more precise. As determined empirically by ascertainment of ancestral polymorphisms and theoretically, the effect of de novo mutation on FST estimation with human whole genome sequence data is statistically negligible. The effect of down-sampling ancestral polymorphisms is also statistically negligible. CONCLUSIONS These results improve and simplify the use and interpretation of FST in studies of population structure.

中文翻译：

---

FST的估计和从头突变的影响。

目的赖特将FST定义为遗传分化的量度。Cockerham在ANOVA框架中基于二元指标开发了FST估算器。在这里，我们解决有关FST估算的2个问题。首先，我们使用双截断正态分布作为二项分布的近似值，基于ANOVA框架推导了FST的新估计量。第二，我们考虑从头突变对FST估计的影响。方法我们通过计算机模拟将估计量与Weir和Cockerham的估计量进行比较。我们将估算器应用于1000个基因组计划的全基因组序列数据。我们使用黑猩猩全基因组序列数据来确定祖先的多态性。结果通过仿真，我们的新估计量比Weir和Cockerham' ，用于比较两个亚群的估计量，系统上更为精确。通过确定祖先多态性从理论上确定，从理论上讲，从头突变对人全基因组序列数据对FST估计的影响在统计上可忽略不计。下采样祖先多态性的影响在统计上也可以忽略不计。结论这些结果改善并简化了FST在人口结构研究中的使用和解释。