Population structure studies and genetic variability in humans

Summary

INTRODUCTION

This short review of human genetic structure deals with genetic distances, display techniques, F-statistics, new types of genetic data, and genetic data other than gene frequencies (migration, isonymy, anthropometry, and pedigrees). This treatment will not be comprehensive, and there are already several other reviews, some of which are more detailed (e.g., Cannings & Cavalli-Sforza, 1973; Fix, 1979; Goodman, 1974; Gower, 1972; Harpending, 1974; Howells, 1973; Jorde, 1980, 1985; Lalouel, 1980; Leslie, 1985; Relethford & Lees, 1982; Roberts, 1975; Smith, 1977; Swedlund, 1980). In addition, applications of many of these methods are contained in the volume by Crawford and Mielke (1982). Since the mathematical equations underlying these methods are readily available in other reviews and in the original papers, most are not repeated here.

GENETIC DISTANCE

Genetic distance measures can be grouped into five broad categories: chi-squared, angular transformation, and gene substitution distances, information measures and non-parametric measures. Each of these will be discussed briefly.

The chi-squared distances include those of Sanghvi (1953), Balakrishnan and Sanghvi (1968), Morton et al (1971), Harpending and Jenkins (1973), Reynolds et al (1983), Steinberg et al (1967), and Kurczynski (1970). All of these involve the calculation of a squared difference between gene frequencies in two populations and the standardisation of this difference. These approaches are most satisfactory when the differences between gene frequencies in subpopulations are not too large (Jorde, 1985).

The angular transformation distances (Edwards, 1971; Edwards & Cavalli-Sforza, 1972) use an arcsine transformation of gene frequencies in order to make the variances of the frequencies independent of the frequency values.