Estimating FST and kinship for arbitrary population structures

F ST and kinship are key parameters often estimated in modern population genetics studies order to quantitatively characterize structure relatedness. Kinship matrices have also become a fundamental quantity used genome-wide association heritability estimation. The most frequently-used estimators of method-of-moments whose accuracies depend strongly on the existence simple underlying forms structure, such as independent subpopulations model non-overlapping, independently evolving subpopulations. However, data sets revealed that these models likely do not hold many populations, including humans. In this work, we analyze behavior presence arbitrarily-complex structures, which results an improved estimation framework specifically designed for arbitrary structures. After generalizing definition structures establishing assessing bias consistency estimators, calculate accuracy existing under characterizing biases challenges unobserved their originally-assumed structure. We then present our new approach, consistently estimates when minimum value dataset is consistently. illustrate using simulated genotypes from admixture model, constructing one-dimensional geographic scenario departs nontrivially model. Our simulations reveal potential severe approaches overcome by framework. This work may significantly improve future analyses rely accurate estimates.