Mateus H. Gouveia

Origin and dynamics of admixture in Brazilians and its effect on the pattern of deleterious mutations

Significance The EPIGEN Brazil Project is the largest Latin-American initiative to study the genomic diversity of admixed populations and its effect on phenotypes. We studied 6,487 Brazilians from three population-based cohorts with different geographic and demographic backgrounds. We identified ancestry components of these populations at a previously unmatched geographic resolution. We broadened our understanding of the African diaspora, the principal destination of which was Brazil, by revealing an African ancestry component that likely derives from the slave trade from Bantu/eastern African populations. In the context of the current debate about how the pattern of deleterious mutations varies between Africans and Europeans, we use whole-genome data to show that continental admixture is the main and complex determinant of the amount of deleterious genotypes in admixed individuals. While South Americans are underrepresented in human genomic diversity studies, Brazil has been a classical model for population genetics studies on admixture. We present the results of the EPIGEN Brazil Initiative, the most comprehensive up-to-date genomic analysis of any Latin-American population. A population-based genome-wide analysis of 6,487 individuals was performed in the context of worldwide genomic diversity to elucidate how ancestry, kinship, and inbreeding interact in three populations with different histories from the Northeast (African ancestry: 50%), Southeast, and South (both with European ancestry >70%) of Brazil. We showed that ancestry-positive assortative mating permeated Brazilian history. We traced European ancestry in the Southeast/South to a wider European/Middle Eastern region with respect to the Northeast, where ancestry seems restricted to Iberia. By developing an approximate Bayesian computation framework, we infer more recent European immigration to the Southeast/South than to the Northeast. Also, the observed low Native-American ancestry (6–8%) was mostly introduced in different regions of Brazil soon after the European Conquest. We broadened our understanding of the African diaspora, the major destination of which was Brazil, by revealing that Brazilians display two within-Africa ancestry components: one associated with non-Bantu/western Africans (more evident in the Northeast and African Americans) and one associated with Bantu/eastern Africans (more present in the Southeast/South). Furthermore, the whole-genome analysis of 30 individuals (42-fold deep coverage) shows that continental admixture rather than local post-Columbian history is the main and complex determinant of the individual amount of deleterious genotypes.

A minimum set of ancestry informative markers for determining admixture proportions in a mixed American population: the Brazilian set

The Brazilian population is considered to be highly admixed. The main contributing ancestral populations were European and African, with Amerindians contributing to a lesser extent. The aims of this study were to provide a resource for determining and quantifying individual continental ancestry using the smallest number of SNPs possible, thus allowing for a cost- and time-efficient strategy for genomic ancestry determination. We identified and validated a minimum set of 192 ancestry informative markers (AIMs) for the genetic ancestry determination of Brazilian populations. These markers were selected on the basis of their distribution throughout the human genome, and their capacity of being genotyped on widely available commercial platforms. We analyzed genotyping data from 6487 individuals belonging to three Brazilian cohorts. Estimates of individual admixture using this 192 AIM panels were highly correlated with estimates using ~370 000 genome-wide SNPs: 91%, 92%, and 74% of, respectively, African, European, and Native American ancestry components. Besides that, 192 AIMs are well distributed among populations from these ancestral continents, allowing greater freedom in future studies with this panel regarding the choice of reference populations. We also observed that genetic ancestry inferred by AIMs provides similar association results to the one obtained using ancestry inferred by genomic data (370 K SNPs) in a simple regression model with rs1426654, related to skin pigmentation, genotypes as dependent variable. In conclusion, these markers can be used to identify and accurately quantify ancestry of Latin Americans or US Hispanics/Latino individuals, in particular in the context of fine-mapping strategies that require the quantification of continental ancestry in thousands of individuals.

Bayesian inferences suggest that Amazon Yunga Natives diverged from Andeans less than 5000 ybp: Implications for South American prehistory

BackgroundArchaeology reports millenary cultural contacts between Peruvian Coast-Andes and the Amazon Yunga, a rainforest transitional region between Andes and Lower Amazonia. To clarify the relationships between cultural and biological evolution of these populations, in particular between Amazon Yungas and Andeans, we used DNA-sequence data, a model-based Bayesian approach and several statistical validations to infer a set of demographic parameters.ResultsWe found that the genetic diversity of the Shimaa (an Amazon Yunga population) is a subset of that of Quechuas from Central-Andes. Using the Isolation-with-Migration population genetics model, we inferred that the Shimaa ancestors were a small subgroup that split less than 5300 years ago (after the development of complex societies) from an ancestral Andean population. After the split, the most plausible scenario compatible with our results is that the ancestors of Shimaas moved toward the Peruvian Amazon Yunga and incorporated the culture and language of some of their neighbors, but not a substantial amount of their genes. We validated our results using Approximate Bayesian Computations, posterior predictive tests and the analysis of pseudo-observed datasets.ConclusionsWe presented a case study in which model-based Bayesian approaches, combined with necessary statistical validations, shed light into the prehistoric demographic relationship between Andeans and a population from the Amazon Yunga. Our results offer a testable model for the peopling of this large transitional environmental region between the Andes and the Lower Amazonia. However, studies on larger samples and involving more populations of these regions are necessary to confirm if the predominant Andean biological origin of the Shimaas is the rule, and not the exception.

Socioeconomic Position, But Not African Genomic Ancestry, Is Associated With Blood Pressure in the Bambui-Epigen (Brazil) Cohort Study of Aging

The study objective is to examine the role of African genome origin on baseline and 11-year blood pressure trajectories in community-based ethnoracially admixed older adults in Brazil. Data come from 1272 participants (aged ≥60 years) of the Bambui cohort study of aging during 11 years of follow-up. Outcome measures were systolic blood pressure, diastolic blood pressure, and hypertension control. Potential confounding variables were demographic characteristics, socioeconomic position (schooling and household income), and health indicators (smoking, sedentary lifestyle, high-density lipoprotein cholesterol, waist circumference, diabetes mellitus, and cardiovascular diseases), including antihypertensive drug use. We used 370 539 single-nucleotide polymorphisms to estimate each individual’s African, European, and Native American trihybrid ancestry proportions. Median African, European, and Native American ancestry were 9.6%, 84.0%, and 5.3%, respectively. Among those with African ancestry, 59.4% came from East and 40.6% from West Africa. Baseline systolic and diastolic blood pressure, controlled hypertension, and their respective trajectories, were not significantly (P>0.05) associated with level (in quintiles) of African genomic ancestry. Similar results were found for West and East African subcontinental origins. Lower schooling level (<4 years versus higher) showed a significant and positive association with systolic blood pressure (Adjusted &bgr;=2.92; 95% confidence interval, 0.85–4.99). Lower monthly household income per capita (<USD 180.00 versus higher) showed an inverse association with hypertension control (&bgr;=−0.35; 95% confidence interval, −0.63 to −0.08, respectively). Our results support the view that favors social and environmental factors as determinants of blood pressure and hypertension control.

Population genetics of immune-related multilocus copy number variation in Native Americans

While multiallelic copy number variation (mCNV) loci are a major component of genomic variation, quantifying the individual copy number of a locus and defining genotypes is challenging. Few methods exist to study how mCNV genetic diversity is apportioned within and between populations (i.e. to define the population genetic structure of mCNV). These inferences are critical in populations with a small effective size, such as Amerindians, that may not fit the Hardy–Weinberg model due to inbreeding, assortative mating, population subdivision, natural selection or a combination of these evolutionary factors. We propose a likelihood-based method that simultaneously infers mCNV allele frequencies and the population structure parameter f, which quantifies the departure of homozygosity from the Hardy–Weinberg expectation. This method is implemented in the freely available software CNVice, which also infers individual genotypes using information from both the population and from trios, if available. We studied the population genetics of five immune-related mCNV loci associated with complex diseases (beta-defensins, CCL3L1/CCL4L1, FCGR3A, FCGR3B and FCGR2C) in 12 traditional Native American populations and found that the population structure parameters inferred for these mCNVs are comparable to but lower than those for single nucleotide polymorphisms studied in the same populations.

Suggestive association between variants in IL1RAPL and asthma symptoms in Latin American children

Several genome-wide association studies have been conducted to investigate the influence of genetic polymorphisms in the development of allergic diseases, but few of them have included the X chromosome. The aim of present study was to perform an X chromosome-wide association study (X-WAS) for asthma symptoms. The study included 1307 children of which 294 were asthma cases. DNA was genotyped using 2.5 HumanOmni Beadchip from Illumina. Statistical analyses were performed in PLINK 1.9, MACH 1.0 and Minimac2. The variant rs12007907 (g.29483892C>A) in IL1RAPL gene was suggestively associated with asthma symptoms in discovery set (odds ratio (OR)=0.49, 95% confidence interval (CI): 0.37–0.67; P=3.33 × 10−6). This result was replicated in the ProAr cohort in men only (OR=0.45, 95% CI: 0.21–0.95; P=0.038). Furthermore, investigating the functional role of the rs12007907 on the production a Th2-type cytokine, IL-13, we found a negative association between the minor allele A with IL-13 production in the discovery set (P=0.044). Gene-based analysis revealed that NUDT10 was the most consistently associated with asthma symptoms in discovery sample. In conclusion, the rs12007907 variant in IL1RAPL gene was negatively associated with asthma and IL-13 production in our study and a sex-specific association was observed in one of the validation samples. It suggests an effect on asthma susceptibility and may explain differences in severe asthma frequency between women and men.

EPIGEN-Brazil Initiative resources: a Latin American imputation panel and the Scientific Workflow

EPIGEN-Brazil is one of the largest Latin American initiatives at the interface of human genomics, public health, and computational biology. Here, we present two resources to address two challenges to the global dissemination of precision medicine and the development of the bioinformatics know-how to support it. To address the underrepresentation of non-European individuals in human genome diversity studies, we present the EPIGEN-5M+1KGP imputation panel-the fusion of the public 1000 Genomes Project (1KGP) Phase 3 imputation panel with haplotypes derived from the EPIGEN-5M data set (a product of the genotyping of 4.3 million SNPs in 265 admixed individuals from the EPIGEN-Brazil Initiative). When we imputed a target SNPs data set (6487 admixed individuals genotyped for 2.2 million SNPs from the EPIGEN-Brazil project) with the EPIGEN-5M+1KGP panel, we gained 140,452 more SNPs in total than when using the 1KGP Phase 3 panel alone and 788,873 additional high confidence SNPs (info score ≥ 0.8). Thus, the major effect of the inclusion of the EPIGEN-5M data set in this new imputation panel is not only to gain more SNPs but also to improve the quality of imputation. To address the lack of transparency and reproducibility of bioinformatics protocols, we present a conceptual Scientific Workflow in the form of a website that models the scientific process (by including publications, flowcharts, masterscripts, documents, and bioinformatics protocols), making it accessible and interactive. Its applicability is shown in the context of the development of our EPIGEN-5M+1KGP imputation panel. The Scientific Workflow also serves as a repository of bioinformatics resources.

Reducing cryptic relatedness in genomic data sets via a central node exclusion algorithm

Cryptic relatedness is a confounding factor in genetic diversity and genetic association studies. Development of strategies to reduce cryptic relatedness in a sample is a crucial step for downstream genetic analyses. This study uses a node selection algorithm, based on network degrees of centrality, to evaluate its applicability and impact on evaluation of genetic diversity and population stratification. 1,036 Guzerá (Bos indicus) females were genotyped using Illumina Bovine SNP50 v2 BeadChip. Four strategies were compared. The first and second strategies consist on a iterative exclusion of most related individuals based on PLINK kinship coefficient (φij) and VanRadens φij, respectively. The third and fourth strategies were based on a node selection algorithm. The fourth strategy, Network G matrix, preserved the larger number of individuals with a better diversity and representation from the initial sample. Determining the most probable number of populations was directly affected by the kinship metric. Network G matrix was the better strategy for reducing relatedness due to producing a larger sample, with more distant individuals, a more similar distribution when compared with the full data set in the MDS plots and keeping a better representation of the population structure. Resampling strategies using VanRadens φij as a relationship metric was better to infer the relationships among individuals. Moreover, the resampling strategies directly impact the genomic inflation values in genomewide association studies. The use of the node selection algorithm also implies better selection of the most central individuals to be removed, providing a more representative sample.

Genomic African and Native American Ancestry and 15-Year Cognitive Trajectory: Bambui Study, Brazil: Genomic ancestry and cognitive trajectory

To investigate the association between African and Native American genomic ancestry and long‐term cognitive trajectories in admixed Brazilians.