Recombination of haplotypes leads to biased estimates of admixture proportions in human populations.

Abstract

A population formed by genetic admixture of two or more source populations may exhibit considerable linkage disequilibrium between genetic loci. In the presence of recombination, this linkage disequilibrium declines with time, a fact that is often ignored when considering haplotypes of closely linked systems Admixture analysis in human populations traditionally uses polymorphic genetic markers. Careful reviews of the subject indicate that the precision and utility of admixture analysis can be greatly enhanced by utilizing genetic systems that demonstrate substantial differences in the allele frequencies of the populations between which genetic admixture has occurred (1-3). Although the traditional polymorphic immunological and biochemical markers have increased our understanding of genetic admixture in many human populations (e.g., American Blacks, Ashkenazi Jews, and Icelanders), the results for many of these studies are not fully concordant with the known ethnohistory (1). This is partly due to the fact that traditional genetic markers do not always exhibit wide differences of allele frequency among the parental populations. This limitation can be partially circumvented by increasing the degree of polymorphism at a given locus with more refined assay techniques, such as isoelectric focusing for the phosphoglucomutase 1, group-specific component (Gc), and other serum protein systems.Greater polymorphism and interpopulation genetic variability are revealed when haplotype frequencies at linked loci are considered. For example, while the numbers of segregating alleles at the HLA-A and -B loci are at least 17 and 27, respectively, the number of AB haplotypes segregating in the American Black population is approximately 146 (4). It may be advantageous to use haplotype frequencies to determine admixture proportions, instead of the allele frequencies at the separate genetic loci. The HLA system may not be the best material for admixture analysis because of the selective implications of genetic variation at the histocompatibility complex (5, 6), but the same considerations apply to several other sets of loci, such as the Rhesus blood group, MNS blood group, and Gm serum gamma-globulin systems. They also apply to linked sets of DNA markers revealed by restriction analysis.Our purpose here is to show that when genetic loci (or restriction sutes) are not absolutely linked and when there is recombination between them, analysis of admixture proportions from haplotype frequencies yields biased values. The main source of this bias is recombination, in consequence of which the haplotype frequencies in a contemporary population will be different from those in the ancestral population from which it is derived. Even where linkage relationships between loci are known, allele frequencies at individual loci may better reflect the admixture history than will haplotype frequencies. When the history of admixture is known, however, ancestral haplotype frequencies may be inferred from their contemporary values. In that case, ha...

Admixture as a tool for finding linked genes and detecting that difference from allelic association between loci.

“…Here, we refer to such gametic association as "mixture disequilibrium" to distinguish it from gametic association between closely linked loci. Such mixture disequilibrium will decay over time, but (2) showed that in the presence of recombination, estimates of admixture from haplotype data may be errorprone, if the genetic assay of the admixed population is not done immediately after the admixture event. There are many instances where human populations have been formed through admixture of the same two stocks of racial groups, yet the degree of admixture varies among the admixed groups (with approximately the same historical depth of admixture).…”

“…Under these assumptions, it is known (2,4) that the mixture disequilibrium between the A and B loci in the admixed population Z, produced by admixture, at generation 0 can be written as [1] where 8A Thus, for given values of 8A and 8B (allele frequency differences between the two parental populations) and mixture Abbreviation: lod, log-likelihood ratio.…”

Review and Evaluation of Methods Correcting for Population Stratification with a Focus on Underlying Statistical Principles

“…As previously discussed, a consequence of PS in association studies is the potential for bias in the estimate of allelic associations due to deviations from the HardyWeinberg equilibrium and the induction of linkage disequilibrium [156,157] . In order for bias due to PS to exist, both the frequency of the marker variant of interest and the background disease prevalence must vary significantly by race/ethnicity [158,159] .…”

Transmission disequilibrium test with discordant sib pairs when parents are available

“…Association studies that depend on LD between markers and disease genes have helped to decipher some genetic basis of differential susceptibility to complex diseases (e.g., Feder et al 1996). Regular association studies, usually, case-control analyses in unrelated cases and controls may suffer inflated type I errors (Chakraborty and Smouse 1988;Lander and Schork 1994;Weir 1996;Spielman and Ewens 1996) that have not been quantified until recently (Deng and Chen 1999;Deng et al 2001a). In addition, population admixture/stratification may mask or reverse true genetic effects in classical association studies (Deng 2001).…”