Abordagens de biologia computacional para o estudo da diversidade genômica dos brasileiros

Abstract

Brazil is the largest and the most populous Latin American country. It has more than 200 millions inhabitants, which are the product of post-Columbian admixture between Native American, European and Africans. Latin Americans are classical models for the studies of the effect of admixture on human populations, but they are underrepresented in modern studies on the human genomic diversity. This Master thesis is part of the EPIGEN-Brasil initiative, the largest Latin American initiative aimed to study the genomic diversity of this part of the world. Two goals of the EPIGEN project are: (i) to identify and quantify for the first time, biogeographic components of ancestry of the Brazilian population at a sub-continental level, (ii) to infer admixture dynamics of the Brazilian population. To achieve these goals, we implemented two computational approaches. The first approach uses complex network theory to identify sets of relatives departing from a matrix of kinship coefficients, and suggest a heuristic methodology to reduce the level of kinship in a populational sample, minimizing the number of individuals to be excluded from the sample. This is a relevant problem because the presence of related individuals generate artifacts in the analysis of biogeographic ancestry, and related individuals have to be retired from the analysis. The second approach develops and implements a new methodology based on Approximate Bayesian Calculation to infer the a posteriori distribution of parameter that characterize the dynamics of the historical process of admixture. This method is applied to the Brazilian population, revealing the signature of more recent European gene flow in Southeast/South Brazil than in Northeast.