From Populations to Clines in Modern Statistical Genetics

© Oxford University Press 2015. All rights reserved. Genome-wide association studies enable alleles carried by individuals in natural populations to be correlated at a continuum of scales, from close to distant relatives across populations. A central problem is to isolate subpopulations or subsets of alleles amongst the billions of possible combinations that make up the genome. Models need to allow for natural selection, genetic drift, and historical factors, like migration. This chapter discusses alternative models and, more particularly, the nature of inferences they require. In some genome-wide association studies, allele frequency differences between cases (affected individuals) and controls (putatively unaffected individuals) are used to identify parts of the genome predisposing to the phenotype of interest. More recent methods, rather than initially grouping individuals into populations, permit analysis in effect to treat each individual as a population, then modelling genome-wide structures from the bottom up by correlating patterns of variation at a given gene locus between pairs of individuals.