Iterative pruning PCA improves resolution of highly structured populations

Assessing the power of principal components and wright's fixation index analyzes applied to reveal the genome-wide genetic differences between herds of Holstein cows

Background

Due to the advent of SNP array technology, a genome-wide analysis of genetic differences between populations and breeds has become possible at a previously unattainable level. The Wright’s fixation index (F_st) and the principal component analysis (PCA) are widely used methods in animal genetics studies. In paper we compared the power of these methods, their complementing each other and which of them is the most powerful.

Results

Comparative analysis of the power Principal Components Analysis (PCA) and F_st were carried out to reveal genetic differences between herds of Holsteinized cows. Totally, 803 BovineSNP50 genotypes of cows from 13 herds were used in current study. Obtained F_st values were in the range of 0.002–0.012 (mean 0.0049) while for rare SNPs with MAF 0.0001–0.005 they were even smaller in the range of 0.001–0.01 (mean 0.0027). Genetic relatedness of the cows in the herds was the cause of such small F_st values. The contribution of rare alleles with MAF 0.0001–0.01 to the F_st values was much less than common alleles and this effect depends on linkage disequilibrium (LD). Despite of substantial change in the MAF spectrum and the number of SNPs we observed small effect size of LD - based pruning on F_st data. PCA analysis confirmed the mutual admixture and small genetic difference between herds. Moreover, PCA analysis of the herds based on the visualization the results of a single eigenvector cannot be used to significantly differentiate herds. Only summed eigenvectors should be used to realize full power of PCA to differentiate small between herds genetic difference. Finally, we presented evidences that the significance of F_st data far exceeds the significance of PCA data when these methods are used to reveal genetic differences between herds.

Conclusions

LD - based pruning had a small effect on findings of F_st and PCA analyzes. Therefore, for weakly structured populations the LD - based pruning is not effective. In addition, our results show that the significance of genetic differences between herds obtained by F_st analysis exceeds the values of PCA. Proposed, to differentiate herds or low structured populations we recommend primarily using the F_st approach and only then PCA.

Ancestry-informative marker (AIM) SNP panel for the Malay population

Ancestry-informative markers (AIMs) can be used to infer the ancestry of an individual to minimize the inaccuracy of self-reported ethnicity in biomedical research. In this study, we describe three methods for selecting AIM SNPs for the Malay population (Malay AIM panel) using different approaches based on pairwise F_ST, informativeness for assignment (I_n), and PCA-correlated SNPs (PCAIMs). These Malay AIM panels were extracted from genotype data stored in SNP arrays hosted by the Malaysian node of the Human Variome Project (MyHVP) and the Singapore Genome Variation Project (SGVP). In particular, genotype data from a total of 165 Malay individuals were analyzed, comprising data on 117 individual genotypes from the Affymetrix SNP-6 SNP array platform and data on 48 individual genotypes from the OMNI 2.5 Illumina SNP array platform. The HapMap phase 3 database (1397 individuals from 11 populations) was used as a reference for comparison with the Malay genotype data. The accuracy of each resulting Malay AIM panel was evaluated using a machine learning "ancestry-predictive model" constructed by using WEKA, a comprehensive machine learning platform written in Java. A total of 1250 SNPs were finally selected, which successfully identified Malay individuals from other world populations with an accuracy of 90%, but the accuracy decreased to 80% using 157 SNPs according to the pairwise F_ST method, while a panel of 200 SNPs selected using I_n and PCAIMs could be used to identify Malay individuals with an accuracy of approximately 80%.

A different view on fine-scale population structure in Western African populations

Due to its long genetic evolutionary history, Africans exhibit more genetic variation than any other population in the world. Their genetic diversity further lends itself to subdivisions of Africans into groups of individuals with a genetic similarity of varying degrees of granularity. It remains challenging to detect fine-scale structure in a computationally efficient and meaningful way. In this paper, we present a proof-of-concept of a novel fine-scale population structure detection tool with Western African samples. These samples consist of 1396 individuals from 25 ethnic groups (two groups are African American descendants). The strategy is based on a recently developed tool called IPCAPS. IPCAPS, or Iterative Pruning to CApture Population Structure, is a genetic divisive clustering strategy that enhances iterative pruning PCA, is robust to outliers and does not require a priori computation of haplotypes. Our strategy identified in total 12 groups and 6 groups were revealed as fine-scale structure detected in the samples from Cameroon, Gambia, Mali, Southwest USA, and Barbados. Our finding helped to explain evolutionary processes in the analyzed West African samples and raise awareness for fine-scale structure resolution when conducting genome-wide association and interaction studies.

IPCAPS: an R package for iterative pruning to capture population structure

Background

Resolving population genetic structure is challenging, especially when dealing with closely related or geographically confined populations. Although Principal Component Analysis (PCA)-based methods and genomic variation with single nucleotide polymorphisms (SNPs) are widely used to describe shared genetic ancestry, improvements can be made especially when fine-scale population structure is the target.

Results

This work presents an R package called IPCAPS, which uses SNP information for resolving possibly fine-scale population structure. The IPCAPS routines are built on the iterative pruning Principal Component Analysis (ipPCA) framework that systematically assigns individuals to genetically similar subgroups. In each iteration, our tool is able to detect and eliminate outliers, hereby avoiding severe misclassification errors.

Conclusions

IPCAPS supports different measurement scales for variables used to identify substructure. Hence, panels of gene expression and methylation data can be accommodated as well. The tool can also be applied in patient sub-phenotyping contexts. IPCAPS is developed in R and is freely available from http://bio3.giga.ulg.ac.be/ipcaps.

Nonparametric approaches for population structure analysis

The analysis of population structure has many applications in medical and population genetic research. Such analysis is used to provide clear insight into the underlying genetic population substructure and is a crucial prerequisite for any analysis of genetic data. The analysis involves grouping individuals into subpopulations based on shared genetic variations. The most widely used markers to study the variation of DNA sequences between populations are single nucleotide polymorphisms. Data preprocessing is a necessary step to assess the quality of the data and to determine which markers or individuals can reasonably be included in the analysis. After preprocessing, several methods can be utilized to uncover population substructure, which can be categorized into two broad approaches: parametric and nonparametric. Parametric approaches use statistical models to infer population structure and assign individuals into subpopulations. However, these approaches suffer from many drawbacks that make them impractical for large datasets. In contrast, nonparametric approaches do not suffer from these drawbacks, making them more viable than parametric approaches for analyzing large datasets. Consequently, nonparametric approaches are increasingly used to reveal population substructure. Thus, this paper reviews and discusses the nonparametric approaches that are available for population structure analysis along with some implications to resolve challenges.

A comparison of DMET Plus microarray and genome-wide technologies by assessing population substructure

OBJECTIVE

The capacity of the Affymetrix drug metabolism enzymes and transporters (DMET) Plus pharmacogenomics genotyping chip to estimate population substructure and cryptic relatedness was evaluated. The results were compared with estimates using genome-wide HapMap data for the same individuals.

METHODS

For 301 unrelated individuals, spanning three continental populations and one admixed population, genotypic data were collected using the Affymetrix DMET Plus microarray. Genome-wide data on these individuals were obtained from HapMap release 3. Population substructure was assessed using Eigenstrat and ADMIXTURE software for both platforms. Cryptic relatedness was explored by inbreeding coefficient estimation. Nonparametric tests were used to determine correlations of the analytical results of the two genotyping platforms.

RESULTS

Principal components analysis identified population substructure for both datasets, with 15.8 and 16.6% of the total variance explained in the first two principal components for DMET Plus and HapMap data, respectively. ADMIXTURE results correctly identified four subpopulations within each dataset. Nonparametric rank correlations indicated significant associations between analyses with an average ρ=0.7272 (P<10) across the three continental populations and ρ=0.4888 for the admixed population. Concordance correlation coefficients (average ρc=0.9693 across all four subpopulations) strongly indicate concordance between ADMIXTURE results. Inbreeding coefficients were slightly inflated (16 individuals>0.15) using DMET Plus data and no cryptic relatedness was indicated using HapMap data. The inflated inbreeding estimation could be because of the limited number of markers provided by DMET as a random sample of 1832 markers from HapMap also yielded inflated estimates of cryptic relatedness (39 individuals>0.15). Furthermore, use of single nucleotide polymorphisms located in genes involved in metabolism and transport may have different allele frequencies in subpopulations than single nucleotide polymorphisms sampled from the whole genome.

CONCLUSION

The DMET Plus pharmacogenomics genotyping chip is effective in quantifying population substructure across the three continental populations and inferring the presence of an admixed population. On the basis of our results, these microarrays offer sufficient depth for covariate adjustment of population substructure in genomic association studies.

HaploPOP: a software that improves population assignment by combining markers into haplotypes

Background

In ecology and forensics, some population assignment techniques use molecular markers to assign individuals to known groups. However, assigning individuals to known populations can be difficult if the level of genetic differentiation among populations is small. Most assignment studies handle independent markers, often by pruning markers in Linkage Disequilibrium (LD), ignoring the information contained in the correlation among markers due to LD.

Results

To improve the accuracy of population assignment, we present an algorithm, implemented in the HaploPOP software, that combines markers into haplotypes, without requiring independence. The algorithm is based on the Gain of Informativeness for Assignment that provides a measure to decide if a pair of markers should be combined into haplotypes, or not, in order to improve assignment. Because complete exploration of all possible solutions for constructing haplotypes is computationally prohibitive, our approach uses a greedy algorithm based on windows of fixed sizes. We evaluate the performance of HaploPOP to assign individuals to populations using a split-validation approach. We investigate both simulated SNPs data and dense genotype data from individuals from Spain and Portugal.

Conclusions

Our results show that constructing haplotypes with HaploPOP can substantially reduce assignment error. The HaploPOP software is freely available as a command-line software at www.ieg.uu.se/Jakobsson/software/HaploPOP/.

Challenges in analysis and interpretation of microsatellite data for population genetic studies

Advancing technologies have facilitated the ever-widening application of genetic markers such as microsatellites into new systems and research questions in biology. In light of the data and experience accumulated from several years of using microsatellites, we present here a literature review that synthesizes the limitations of microsatellites in population genetic studies. With a focus on population structure, we review the widely used fixation (F ST) statistics and Bayesian clustering algorithms and find that the former can be confusing and problematic for microsatellites and that the latter may be confounded by complex population models and lack power in certain cases. Clustering, multivariate analyses, and diversity-based statistics are increasingly being applied to infer population structure, but in some instances these methods lack formalization with microsatellites. Migration-specific methods perform well only under narrow constraints. We also examine the use of microsatellites for inferring effective population size, changes in population size, and deeper demographic history, and find that these methods are untested and/or highly context-dependent. Overall, each method possesses important weaknesses for use with microsatellites, and there are significant constraints on inferences commonly made using microsatellite markers in the areas of population structure, admixture, and effective population size. To ameliorate and better understand these constraints, researchers are encouraged to analyze simulated datasets both prior to and following data collection and analysis, the latter of which is formalized within the approximate Bayesian computation framework. We also examine trends in the literature and show that microsatellites continue to be widely used, especially in non-human subject areas. This review assists with study design and molecular marker selection, facilitates sound interpretation of microsatellite data while fostering respect for their practical limitations, and identifies lessons that could be applied toward emerging markers and high-throughput technologies in population genetics.

iNJclust: Iterative Neighbor-Joining Tree Clustering Framework for Inferring Population Structure

Understanding genetic differences among populations is one of the most important issues in population genetics. Genetic variations, e.g., single nucleotide polymorphisms, are used to characterize commonality and difference of individuals from various populations. This paper presents an efficient graph-based clustering framework which operates iteratively on the Neighbor-Joining (NJ) tree called the iNJclust algorithm. The framework uses well-known genetic measurements, namely the allele-sharing distance, the neighbor-joining tree, and the fixation index. The behavior of the fixation index is utilized in the algorithm's stopping criterion. The algorithm provides an estimated number of populations, individual assignments, and relationships between populations as outputs. The clustering result is reported in the form of a binary tree, whose terminal nodes represent the final inferred populations and the tree structure preserves the genetic relationships among them. The clustering performance and the robustness of the proposed algorithm are tested extensively using simulated and real data sets from bovine, sheep, and human populations. The result indicates that the number of populations within each data set is reasonably estimated, the individual assignment is robust, and the structure of the inferred population tree corresponds to the intrinsic relationships among populations within the data.

Insight into the peopling of Mainland Southeast Asia from Thai population genetic structure

There is considerable ethno-linguistic and genetic variation among human populations in Asia, although tracing the origins of this diversity is complicated by migration events. Thailand is at the center of Mainland Southeast Asia (MSEA), a region within Asia that has not been extensively studied. Genetic substructure may exist in the Thai population, since waves of migration from southern China throughout its recent history may have contributed to substantial gene flow. Autosomal SNP data were collated for 438,503 markers from 992 Thai individuals. Using the available self-reported regional origin, four Thai subpopulations genetically distinct from each other and from other Asian populations were resolved by Neighbor-Joining analysis using a 41,569 marker subset. Using an independent Principal Components-based unsupervised clustering approach, four major MSEA subpopulations were resolved in which regional bias was apparent. A major ancestry component was common to these MSEA subpopulations and distinguishes them from other Asian subpopulations. On the other hand, these MSEA subpopulations were admixed with other ancestries, in particular one shared with Chinese. Subpopulation clustering using only Thai individuals and the complete marker set resolved four subpopulations, which are distributed differently across Thailand. A Sino-Thai subpopulation was concentrated in the Central region of Thailand, although this constituted a minority in an otherwise diverse region. Among the most highly differentiated markers which distinguish the Thai subpopulations, several map to regions known to affect phenotypic traits such as skin pigmentation and susceptibility to common diseases. The subpopulation patterns elucidated have important implications for evolutionary and medical genetics. The subpopulation structure within Thailand may reflect the contributions of different migrants throughout the history of MSEA. The information will also be important for genetic association studies to account for population-structure confounding effects.

Softwares and methods for estimating genetic ancestry in human populations

The estimation of genetic ancestry in human populations has important applications in medical genetic studies. Genetic ancestry is used to control for population stratification in genetic association studies, and is used to understand the genetic basis for ethnic differences in disease susceptibility. In this review, we present an overview of genetic ancestry estimation in human disease studies, followed by a review of popular softwares and methods used for this estimation.

NetView: a high-definition network-visualization approach to detect fine-scale population structures from genome-wide patterns of variation

High-throughput sequencing and single nucleotide polymorphism (SNP) genotyping can be used to infer complex population structures. Fine-scale population structure analysis tracing individual ancestry remains one of the major challenges. Based on network theory and recent advances in SNP chip technology, we investigated an unsupervised network clustering method called Super Paramagnetic Clustering (Spc). When applied to whole-genome marker data it identifies the natural divisions of groups of individuals into population clusters without use of prior ancestry information. Furthermore, we optimised an analysis pipeline called NetView, a high-definition network visualization, starting with computation of genetic distance, followed clustering using Spc and finally visualization of clusters with Cytoscape. We compared NetView against commonly used methodologies including Principal Component Analyses (PCA) and a model-based algorithm, Admixture, on whole-genome-wide SNP data derived from three previously described data sets: simulated (2.5 million SNPs, 5 populations), human (1.4 million SNPs, 11 populations) and cattle (32,653 SNPs, 19 populations). We demonstrate that individuals can be effectively allocated to their correct population whilst simultaneously revealing fine-scale structure within the populations. Analyzing the human HapMap populations, we identified unexpected genetic relatedness among individuals, and population stratification within the Indian, African and Mexican samples. In the cattle data set, we correctly assigned all individuals to their respective breeds and detected fine-scale population sub-structures reflecting different sample origins and phenotypes. The NetView pipeline is computationally extremely efficient and can be easily applied on large-scale genome-wide data sets to assign individuals to particular populations and to reproduce fine-scale population structures without prior knowledge of individual ancestry. NetView can be used on any data from which a genetic relationship/distance between individuals can be calculated.

Population identification using genetic data

A large number of algorithms have been developed to classify individuals into discrete populations using genetic data. Recent results show that the information used by both model-based clustering methods and principal components analysis can be summarized by a matrix of pairwise similarity measures between individuals. Similarity matrices have been constructed in a number of ways, usually treating markers as independent but differing in the weighting given to polymorphisms of different frequencies. Additionally, methods are now being developed that take linkage into account. We review several such matrices and evaluate their information content. A two-stage approach for population identification is to first construct a similarity matrix and then perform clustering. We review a range of common clustering algorithms and evaluate their performance through a simulation study. The clustering step can be performed either on the matrix or by first using a dimension-reduction technique; we find that the latter approach substantially improves the performance of most algorithms. Based on these results, we describe the population structure signal contained in each similarity matrix and find that accounting for linkage leads to significant improvements for sequence data. We also perform a comparison on real data, where we find that population genetics models outperform generic clustering approaches, particularly with regard to robustness for features such as relatedness between individuals.

Study of large and highly stratified population datasets by combining iterative pruning principal component analysis and structure

Background

The ever increasing sizes of population genetic datasets pose great challenges for population structure analysis. The Tracy-Widom (TW) statistical test is widely used for detecting structure. However, it has not been adequately investigated whether the TW statistic is susceptible to type I error, especially in large, complex datasets. Non-parametric, Principal Component Analysis (PCA) based methods for resolving structure have been developed which rely on the TW test. Although PCA-based methods can resolve structure, they cannot infer ancestry. Model-based methods are still needed for ancestry analysis, but they are not suitable for large datasets. We propose a new structure analysis framework for large datasets. This includes a new heuristic for detecting structure and incorporation of the structure patterns inferred by a PCA method to complement STRUCTURE analysis.

Results

A new heuristic called EigenDev for detecting population structure is presented. When tested on simulated data, this heuristic is robust to sample size. In contrast, the TW statistic was found to be susceptible to type I error, especially for large population samples. EigenDev is thus better-suited for analysis of large datasets containing many individuals, in which spurious patterns are likely to exist and could be incorrectly interpreted as population stratification. EigenDev was applied to the iterative pruning PCA (ipPCA) method, which resolves the underlying subpopulations. This subpopulation information was used to supervise STRUCTURE analysis to infer patterns of ancestry at an unprecedented level of resolution. To validate the new approach, a bovine and a large human genetic dataset (3945 individuals) were analyzed. We found new ancestry patterns consistent with the subpopulations resolved by ipPCA.

Conclusions

The EigenDev heuristic is robust to sampling and is thus superior for detecting structure in large datasets. The application of EigenDev to the ipPCA algorithm improves the estimation of the number of subpopulations and the individual assignment accuracy, especially for very large and complex datasets. Furthermore, we have demonstrated that the structure resolved by this approach complements parametric analysis, allowing a much more comprehensive account of population structure. The new version of the ipPCA software with EigenDev incorporated can be downloaded from http://www4a.biotec.or.th/GI/tools/ippca.