The cumulative effect of genetic markers on classification performance: Insights from simple models

Janus-faced race: Is race biological, social, or mythical?

As belief in the reality of race as a biological category among U.S. anthropologists has fallen, belief in the reality of race as a social category has risen in its place. The view that race simply does not exist-that it is a myth-is treated with suspicion. While racial classification is linked to many of the worst evils of recent history, it is now widely believed to be necessary to fight back against racism. In this article, I argue that race is indeed a biological fiction, but I critique the claim that race is socially real. I defend a form of anti-realist reconstructionism about race, which says that there are no races, only racialized groups-groups mistakenly believed to be races. I argue that this is the most attractive position about race from a metaphysical perspective, and that it is also the position most conductive to public understanding and social justice.

A Meta-Analysis of the "Erasing Race" Effect in the United States and Some Theoretical Considerations

The "erasing race" effect is the reduction of the salience of "race" as an alliance cue when recalling coalition membership, once more accurate information about coalition structure is presented. We conducted a random-effects model meta-analysis of this effect using five United States studies (containing nine independent effect sizes). The effect was found (ρ = 0.137, K = 9, 95% CI = 0.085 to 0.188). However, no decline effect or moderation effects were found (a "decline effect" in this context would be a decrease in the effect size over time). Furthermore, we found little evidence of publication bias. Synthetically correcting the effect size for bias stemming from the use of an older method for calculating error base rates reduced the magnitude of the effect, but the it remained significant. Taken together, these findings indicate that the "erasing race" effect generalizes quite well across experimental contexts and would, therefore, appear to be quite robust. We reinterpret the theoretical basis for these effects in line with Brunswikian evolutionary-developmental theory and present a series of predictions to guide future research in this area.

New perspectives on multilocus ancestry informativeness

We present an axiomatic approach for multilocus informativeness measures for determining the amount of information that a set of polymorphic genetic markers provides about individual ancestry. We then reveal several surprising properties of a decision-theoretic based measure that is consistent with the set of proposed criteria for multilocus informativeness. In particular, these properties highlight the interplay between information originating from population priors and the information extractable from the population genetic variants. This analysis then reveals a certain deficiency of mutual information based multilocus informativeness measures when such population priors are incorporated. Finally, we analyse and quantify the inevitable inherent decrease in informativeness due to learning from finite population samples.

From typical sequences to typical genotypes

We demonstrate an application of a core notion of information theory, typical sequences and their related properties, to analysis of population genetic data. Based on the asymptotic equipartition property (AEP) for nonstationary discrete-time sources producing independent symbols, we introduce the concepts of typical genotypes and population entropy and cross entropy rate. We analyze three perspectives on typical genotypes: a set perspective on the interplay of typical sets of genotypes from two populations, a geometric perspective on their structure in high dimensional space, and a statistical learning perspective on the prospects of constructing typical-set based classifiers. In particular, we show that such classifiers have a surprising resilience to noise originating from small population samples, and highlight the potential for further links between inference and communication.

On the Apportionment of Population Structure

Measures of population differentiation, such as F_ST, are traditionally derived from the partition of diversity within and between populations. However, the emergence of population clusters from multilocus analysis is a function of genetic structure (departures from panmixia) rather than of diversity. If the populations are close to panmixia, slight differences between the mean pairwise distance within and between populations (low F_ST) can manifest as strong separation between the populations, thus population clusters are often evident even when the vast majority of diversity is partitioned within populations rather than between them. For any given F_ST value, clusters can be tighter (more panmictic) or looser (more stratified), and in this respect higher F_ST does not always imply stronger differentiation. In this study we propose a measure for the partition of structure, denoted E_ST, which is more consistent with results from clustering schemes. Crucially, our measure is based on a statistic of the data that is a good measure of internal structure, mimicking the information extracted by unsupervised clustering or dimensionality reduction schemes. To assess the utility of our metric, we ranked various human (HGDP) population pairs based on F_ST and E_ST and found substantial differences in ranking order. E_ST ranking seems more consistent with population clustering and classification and possibly with geographic distance between populations. Thus, E_ST may at times outperform F_ST in identifying evolutionary significant differentiation.

Implications of the apportionment of human genetic diversity for the apportionment of human phenotypic diversity

Researchers in many fields have considered the meaning of two results about genetic variation for concepts of "race." First, at most genetic loci, apportionments of human genetic diversity find that worldwide populations are genetically similar. Second, when multiple genetic loci are examined, it is possible to distinguish people with ancestry from different geographical regions. These two results raise an important question about human phenotypic diversity: To what extent do populations typically differ on phenotypes determined by multiple genetic loci? It might be expected that such phenotypes follow the pattern of similarity observed at individual loci. Alternatively, because they have a multilocus genetic architecture, they might follow the pattern of greater differentiation suggested by multilocus ancestry inference. To address the question, we extend a well-known classification model of Edwards (2003) by adding a selectively neutral quantitative trait. Using the extended model, we show, in line with previous work in quantitative genetics, that regardless of how many genetic loci influence the trait, one neutral trait is approximately as informative about ancestry as a single genetic locus. The results support the relevance of single-locus genetic-diversity partitioning for predictions about phenotypic diversity.

Two complementary perspectives on inter-individual genetic distance

This paper examines population structure through the prism of pairwise genetic distances. Two complementary perspectives, framed as two simple questions, are explored: Q1: What is the probability that a random pair of individuals from the same local population is more genetically dissimilar than a random pair from two distinct populations? Q2: On average, how genetically different are two individuals from the same local population, in comparison with two individuals chosen from any two distinct populations? Models are developed to provide quantitative answers for the two questions, given allele frequencies across any number of markers from two diploid populations. The probability from Q1 is shown to drop to zero with increasing number of genetic markers even for very closely-related populations and rare alleles. The average genetic dissimilarity of two individuals from distinct populations diverges from the average dissimilarity of two individuals from the same population by a percentage dependent on estimates of population differentiation. This perspective also suggests a measure of population distance based on the intuitive notion of pairwise genetic distance, along with a simple method of estimation. Results from recent empirical research on inter-individual genetic distance in human populations are analyzed in the context of the theoretical framework.