A genealogical interpretation of principal components analysis.
McVean G. · 2009
Read A genealogical interpretation of principal components analysis. by McVean G. (2009) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc2757795.txt.
Introduction
McVean gave in 2009 a genealogical interpretation of principal components analysis, a statistical method widely used to reveal structure in genetic variation across places and ethnic groups. The method projects each sample onto main axes, but what those projections mean in terms of migration, isolation, and admixture was poorly understood. Practitioners often read history into the plots without a formal basis.
He shows that for SNP data, a sample's projection can be derived directly from average coalescent times, the times back to common ancestors, between pairs of genomes. That links the plot to Wright's Fst, a classic measure of population split. SNP ascertainment, how variants were chosen, distorts the projection in a largely simple and predictable way, which sharpens empirical inference.
Catalog record
| CREATOR | McVean G. |
|---|---|
| YEAR | 2009 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Genealogical Interpretation of Principal Components Analysis By Gil McVean, Molly Przeworski Published in: PLoS Genetics DOI: 10.1371/journal.pgen.1000686 ABSTRACT Principal components analysis, PCA, is a statistical method commonly used in population genetics to identify structure in the distribution of genetic variation across geographical location and ethnic background. However, while the method is often used to inform about historical demographic processes, little is known about the relationship between fundamental demographic parameters and the projection of samples onto the primary axes. Here I show that for SNP data the projection of samples onto the principal components can be obtained directly from considering the average coalescent times between pairs of haploid genomes. The result provides a framework for interpreting PCA projections in terms of underlying processes, including migration, geographical isolation, and admixture. I also demonstrate a link between PCA and Wright's fst and show that SNP ascertainment has a largely simple and predictable effect on the projection of samples. Using examples from human genetics, I discuss the application of these results to empirical data and the implications for inference. INTRODUCTION The distribution of genetic variation across geographical location and ethnic background provides a rich source of information about the historical demographic events and processes experienced by a species. However, while colonization, isolation, migration and admixture all lead to a structuring of genetic variation, in which groups of individuals show greater or lesser relatedness to other groups, making inferences about the nature and timing of such processes is notoriously difficult. There are three key problems. First, there are many different processes that one might want to consider as explanations for patterns of structure in empirical data and efficient inference, even under simple models can be difficult. Second, different processes can lead to similar patterns of structure. For example, equilibrium models of restricted migration can give similar patterns of differentiation to non-equilibrium models of population splitting events (at least in terms of some data summaries such as Wright's ). Third, any species is likely to have experienced many different demographic events and processes in its history and their superposition leads to complex patterns of genetic variability. Consequently, while there is a long history of estimating parameters of demographic models from patterns of genetic variation, such models are often highly simplistic and restricted to a subset of possible explanations. An alternative approach to directly fitting models is to use dimension-reduction and data summary techniques to identify key components of the structure within the data in a model-free manner. Perhaps the most widely used technique, and the most important from a historical perspective, is principal components analysis (PCA). Technical descriptions of PCA can be found elsewhere, however, its key feature is that it can be used to project samples onto a series of orthogonal axes, each of which is made up of a linear combination of allelic or genotypic values across SNPs or other types of variant. These axes are chosen such that the projection of samples along the first axis (or first principal component) explains the greatest possible variance in the data among all possible axes. Likewise, projection of samples onto the second axis maximizes the variance for all possibles axes perpendicular to the first and so on for the subsequent components. Typically, the positions of samples along the first two or three axes are presented, although methods for obtaining the statistical significance of any given axis have been developed [1]. Beyond being non-parametric, PCA has many attractive properties including computational speed, the ability to identify structure caused by diverse processes and its ability to group or separate samples in a striking visual manner; for example, see [2]. PCA has also become widespread in the analysis of disease-association studies where the inclusion of the locations of samples on a limited number of axes as covariates can be used in an attempt to control for population stratification [3]. Although PCA is explicitly a non-parametric data summary, it is nevertheless attractive to attempt to use the projections to make inferences about underlying events and processes. For example, dispersion of sample projections along a line is thought to be diagnostic of the samples being admixed between the two populations at the ends of the line, though these need not always be present [1], while correlations between principal components and geographical axes have been interpreted as evidence for waves of migration [4],[5]. However, while simulation studies have shown that such patterns do occur when the inferred process has acted [1],[6], they can also be caused by other processes or even statistical artefacts. For example, clines in principal components result not just from waves of expansion, but also recurrent bottlenecks, admixture and equilibrium models of spatial structure [6]–[11]. In this paper I develop a framework for understanding how PCA relates to underlying processes and events. I show that the expected location of samples on the principal components can, for single nucleotide polymorphism (SNP) data, be predicted directly from the pairwise coalescence times between samples. Because it is often relatively easy to obtain analytical or numerical solutions to expected coalescence times under explicit population genetics models, it is also possible to obtain expressions for the PCA projections of samples under diverse scenarios, including island models, models with isolation and founder events and historical admixture. The result also highlights some key limitations of PCA. For example, it follows that PCA cannot be used to distinguish between models that lead to the same mean coalescence times (for example models with migration or isolation). Furthermore, PCA projections are strongly influenced by uneven sampling. Using examples from human genetics I discuss the implications of these results for making inferences from PCA of genetic variation data. RESULTS PCA describes structure in the matrix of pairwise coalescence times In this section I provide a brief summary of how PCA is carried out and describe the key result concerning the relationship between PCA and average coalescence time. In what follows I assume that haploid individuals have been sequenced with complete accuracy (diploid samples and the influence of SNP ascertainment will be discussed later). The only polymorphisms present are biallelic SNPs that are the result of a single historical mutation. Let be the allelic state for individual i at locus s (here I assume that the ancestral allele is defined as 0 and the derived allele as 1, however the following also applies for any coding, for example where the minor allele is coded as 1). After removing monomorphic sites the data, , consist of an binary matrix ( is the number of SNPs). In PCA, the first step is to zero-centre the data, so as to create a new matrix, X, where(1)At this stage, the data rows are often normalized so as to have equal variance, however, it is assumed that this is not the case (in practice normalization has little effect for SNP data, though will tend to up-weight the influence of rare variants). Each individual sample can be thought of as representing a point in L-dimensional space, where each dimension (or axis) represents a single SNP. The goal of PCA is to find a new set of orthogonal axes (the principal components), each of which is made up from a linear combination of the original axes, such