No description available.

<p>In this manuscript I seek to demonstrate the importance of gene-environment interactions in cardiovascular disease. This manuscript contains five studies each of which contributes to our understanding of the joint impact of genetic variation and environmental exposures to cardiovascular disease: a candidate gene study for gene-smoking interactions associated with early-onset coronary artery disease, an epidemiology study of the association between traffic-related air pollution and cardiovascular disease, a Genome-Wide Interaction Study for gene-by-traffic related air pollution interactions associated with peripheral arterial disease, a Genome-Wide Interaction Study for gene-by-traffic related air pollution interactions on coronary atherosclerosis burden, and a method for analyzing associations between high-dimensional genomics datasets.</p><p> Smoking is a strong risk factors for coronary artery disease, and may play a causative role in the incidence of coronary artery disease. Smoking had been implicated as a reason for heterogeneity observed in associations between genetic variants on chromosome three and coronary artery disease. I used a family-based early-onset coronary artery disease cohort (GENECARD) to study gene-smoking interactions. I also used data from the three independent cohorts to perform a meta-analysis of gene-smoking interactions focusing on the KALRN gene and Rho-GTPase pathway. I found significant evidence for gene-smoking interactions associations involving variants in KALRN and other Rho-GTPase pathway genes on chromosome 3. </p><p> Though the estimated increase in incident cardiovascular disease or cardiovascular events due to air pollution exposure is modest at 3-5%, the ubiquitous nature of air pollution exposures means it has a substantial population-level impact on cardiovascular disease. Historically genome-wide interaction studies with air pollution have not yielded genome-wide significant interactions, however by implementing statistical tools novel to this field I have discovered significant interactions between genetic variants and traffic-related air pollution that are associated with cardiovascular diseases. </p><p> I studied interactions associated with peripheral arterial disease and the number of diseased coronary vessels (an indicator for coronary artery disease burden) using race-stratified cohort study designs. With peripheral arterial disease I observed that variants in both BMP8A and BMP2 showed evidence for interactions in both European-American and African-American cohorts. In BMP8A I uncovered the first genome-wide significant interaction with air pollution associated with cardiovascular disease. BMP2 gene expression is upregulated after exposure to black carbon, a major component of diesel exhaust, and coding variants within this gene showed evidence for interaction. With the number of diseased coronary vessels I observed that variants in PIGR showed significant evidence for involvement in gene-traffic related air pollution interactions. I observed that coding variation within PIGR was associated with coronary artery disease burden in a gene-by-traffic related air pollution interaction model. As PIGR is involved in the immune response it represents a strong candidate gene discovered via an unbiased genome-wide scan.</p><p> The use of high dimensional data to study chronic disease is becoming commonplace. In order to properly analyze high-dimensional data without suffering from high false-discovery rate penalties, the data is often summarized in a way that takes advantage of the correlation structure. Two common approaches for this are principal components analysis and canonical correlation analysis. However neither of these approaches are appropriate when one preferentially desires to preserve structure within the data. To address this shortcoming I developed constrained canonical correlation analysis (cCCA). With cCCA one can evaluate the correlation between two high dimensional datasets while preferentially preserving structure in one of the datasets. This has uses when studying multi-variate outcomes such as cardiovascular disease using multi-variate predictors such as air pollution. Additionally cCCA can be used to create endophenotype factors that specifically explain the variation within a high-dimensional set of predictors (such as gene expression or metabolomics data) with respect to potential endophenotypes for cardiovascular disease, such as cholesterol measures.</p> / Dissertation

Equine protozoal myeloencephalitis (EPM) is a progressive neurological disease of horses caused by Sarcocystis neurona. Two projects were conducted to identify factors involved in the development of EPM. The first study explored a possible genetic susceptibility to EPM by attempting a genome-wide association study (GWAS) on formalin-fixed, paraffin-embedded (FFPE) tissue from 24 definitively-positive EPM horses. DNA extracted from tissues older than 14 months was inadequate for SNP analysis on the Illumina Equine SNP50 BeadChip probably due to degradation and formalin cross-linking. Results were inconclusive as analysis was not possible with the small sample set. The second study evaluated an artificial infection method in creating a reliable equine EPM model. Five horses were injected intravenously at 4 time points with autologous blood incubated with 1,000,000S. neurona merozoites. Challenged horses progressively developed mild to moderate clinical signs and had detectable S. neurona serum antibodies on day 42 post challenge. Horses appeared to have produced a Th1 immune response and cleared the infection by the conclusion of the study on day 89. No histopathological evidence of S. neurona infection was found within central nervous system tissue. This artificial infection method was not effective in replicating the severe clinical EPM seen in natural infections.

Thesis (Ph. D.)--University of Texas Health Science Center at Houston, School of Public Health, 2007. / Source: Dissertation Abstracts International, Volume: 68-10, Section: B, page: 6419. Advisers: Christopher I. Amos; Ralph F. Frankowski. Includes bibliographical references.

Nontyphoidal Salmonella (NTS) causes invasive, and frequently fatal, disease in African children. The burden of disease secondary to NTS reflects inadequacy of Salmonella-control strategies in Africa, with expanding antibiotic resistance, and no licensed anti-NTS vaccine. The delivery of improved interventions to prevent, diagnose, and treat invasive NTS (iNTS) infection, will be facilitated by an improved understanding of the biological determinants of susceptibility to iNTS, including host genetic factors. To identify host genetic determinants of iNTS disease, we performed a GWAS and replication analysis of NTS bacteraemia in African children. This analysis identified and validated a common genetic variant in STAT4 associated with increased iNTS risk. To characterise the function of the NTS-associated STAT4 variant, we utilised a genotype-selectable bioresource of healthy European adults and samples from African children with iNTS disease. In these experiments, the risk genotype at STAT4 is associated with reduced STAT4 RNA expression in stimulated leukocytes, and reduced IFN&gamma; production in both ex vivo stimulated natural killer cells and in the serum of African children with acute NTS bacteraemia. To validate genetic variation suggestively associated with NTS bacteraemia in the GWAS, NTS-associated loci with evidence of regulatory function were prioritised for functional characterisation. Using in vitro models of intracellular Salmonella infection and RNA interference, I characterise the role of a candidate NTS-susceptibility determinant, EVI5L, in Salmonella infections. Finally, applying a pathway enrichment analysis to the NTS bacteraemia GWAS demonstrated that NTS-associated genetic variation in African children is enriched for methionine salvage enzymes. I further investigate the potential for host-pathogen interaction in this pathway, generating and characterising Salmonella mutants deficient in methionine metabolism. Taken together, these data represent the first unbiased assessment of genetic susceptibility to iNTS disease in unselected populations. These results have important implications for the design of Salmonella-control strategies for use in Africa.

RNA interference (RNAi) is a natural mechanism of cellular defence against exogenous double stranded RNA (dsRNA). The discovery of small dsRNA molecules which can be processed by the RNAi pathway in mammalian cells was one of the key advances in the study of functional genomics. These molecules can be designed to downregulate the expression of specific genes. Collections or libraries of dsRNA molecules targeting an extensive number of genes are now available. Using these libraries, numerous studies have implemented high-throughput screens for the study of molecular effectors of numerous phenotypes. The process of designing an RNAi screen requires the consideration of several critical factors during both the experimental and analysis phases. The experimental screen should aim to reproduce the biological phenomenon studied as closely as possible by choosing an adequate model and screening conditions. Phenotype evaluation and assessment of knockdown effects need careful consideration. The results obtained from large-scale RNAi screens are often complex. An analysis pipeline should be implemented which integrates the biological basis of the phenomenon and facilitates the interpretation of the data. This project designed and implemented an unbiased shRNA screen in two in vitro models relevant to carcinogenesis and diabetes. The first screen implemented used a model of neighbour suppression to study the molecular effectors of the response in tumorigenic cells to growth suppression cues from the surrounding tissue, a cellular interaction relevant in early tumorigenesis. The second screen studied two phenotypes relevant to diabetes: proliferation and resistance to lipotoxicity of β-cells in a reversibly immortalised  cell line. An integrative analysis pipeline was also developed to apply network biology and functional enrichment analysis methods for the interpretation of the data obtained from both screens.

Submitted by Samuel Wallace Boer dos Santos (samuel_wallace_eu@hotmail.com) on 2018-10-05T12:41:08Z No. of bitstreams: 1 Dissertação (compacted).pdf: 1033432 bytes, checksum: 72856099afe05132396d596489c6971b (MD5) / Approved for entry into archive by Neli Silvia Pereira null (nelisps@fcav.unesp.br) on 2018-10-08T17:47:46Z (GMT) No. of bitstreams: 1 santos_swb_me_jabo.pdf: 1033432 bytes, checksum: 72856099afe05132396d596489c6971b (MD5) / Made available in DSpace on 2018-10-08T17:47:46Z (GMT). No. of bitstreams: 1 santos_swb_me_jabo.pdf: 1033432 bytes, checksum: 72856099afe05132396d596489c6971b (MD5) Previous issue date: 2018-07-31 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / Características de eficiência alimentar estão diretamente associadas com a lucratividade e sustentabilidade da bovinocultura de corte. Conversão alimentar, consumo alimentar residual, consumo de matéria seca, eficiência alimentar e ganho em peso, são características importantes para a seleção de animais mais eficientes dentro de um sistema de produção, porém, com exceção do ganho em peso, as demais não vêm sendo consideradas como critérios de seleção devido à dificuldade de obtenção de fenótipos para as mesmas. Com o avanço nas tecnologias de genotipagem e sequenciamento, foram desenvolvidos chips de alta densidade de marcadores do tipo SNP (Single Nucleotide Polymorphism) espalhados pelo genoma. Estas informações moleculares vêm sendo utilizadas em estudos de associação genômica ampla (GWAS) e de seleção genômica (SG). Basicamente, o GWAS permite a identificação de variações genéticas de maior efeito sobre a expressão fenotípica de características de interesse, enquanto a SG visa a predição do valor genômico direto dos candidatos à seleção utilizando apenas a informação molecular, o que tem revolucionado o melhoramento genético por proporcionar a diminuição do intervalo de geração e o aumento da acurácia de predição dos valores genéticos dos animais. Assim sendo, os objetivos do presente trabalho foram: 1) encontrar regiões cromossômicas de maior efeito sobre características de eficiência alimentar em animais Nelore provenientes de dois programas de melhoramento (Instituto de Zootecnia - IZ e Nelore Qualitas), visando encontrar possíveis diferenças/semelhanças entre as populações; 2) avaliar a existência de genes candidatos comum às populações; e 3) avaliar a possibilidade e os benefícios de combinar estas duas populações Nelore em estudos de seleção genômica. Foram utilizadas informações fenotípicas e genotípicas de 1.137 animais do IZ e 817 animais do Qualitas. Os animais foram genotipados com painel de alta densidade (Illumina BovineHD chip) ou tiveram seus genótipos imputados para HD através do software FImpute. Após o controle de qualidade dos genótipos, permaneceram para análise 408.161 SNPs para o IZ e 428.621 SNPs para o Qualitas. O GWAS foi realizado para cada população individualmente, considerando a metodologia GBLUP. Modelos unicaracterísticos foram empregados nas análises, incluindo, além dos efeitos aleatórios de animal e resíduo, os efeitos sistemáticos de grupos de contemporâneos (GC), os quais foram definidos como: sexo, ano de nascimento e instalação (IZ) e ano do teste e baia (Qualitas). Para o IZ também foram incluídos, para todas as características, os efeitos fixos de mês de nascimento, e, como covariáveis, idade do animal (linear), idade da mãe (linear e quadrática) e os dois primeiros componentes principais (obtidos a partir da matriz G). O efeito quadrático da idade do animal foi incluído no modelo apenas para o consumo de matéria seca e ganho médio diário. Para o Qualitas, foi considerado, para todas as características, o efeito linear da idade do animal como covariável. No GWAS, foram encontradas algumas regiões cromossômicas de maior efeito para cada característica nas duas populações, porém, não foram encontradas regiões em comum. No estudo de seleção genômica (SG), foram utilizados dez diferentes abordagens e esquemas envolvendo as duas populações para comparar a acurácia de predição. Em geral, a combinação das populações pode gerar benefícios para a seleção genômica, porém, tais benefícios dependem da característica e do esquema de validação. / Feed efficiency traits are directly associated with the profitability and sustainability of beef cattle. Feed conversion rate, residual feed intake, dry matter intake, feed efficiency and average daily gain are important traits for the selection of more efficiency animals within a production system, but, except for weight gain, the others have not been considered as selection criteria due to the difficulty of obtaining phenotypes. With the advance in genotyping and sequencing technologies, high density chips of SNP (Single Nucleotide Polymorphism) have been developed. This molecular information has been used in genome-wide association (GWAS) and genomic selection (GS) studies. Basically, GWAS allows the identification of genetic variations with major effects on the phenotypic expression of traits of interest, while SG aims at the prediction of direct genomic value for the selection candidates using only their molecular information, which has revolutionized the animal breeding by providing a decrease in generation interval and increases in the prediction accuracies of breeding values. Thus, the objectives of the present study were to: 1) identify chromosomal regions with major effects on feed efficiency traits in animals from two Nellore breeding programs (Instituto de Zootecnia and Nellore Qualitas), in order to find possible differences/similarities between the populations; 2) evaluate the existence of candidate genes in common to populations; and 3) evaluate the possibility and benefits of combining these two Nellore populations in genomic selection studies. Phenotypic and genotypic information of 1,137 animals from IZ and 817 from Qualitas were used. The animals were genotyped with high density panel (Illumina BovineHD chip) or had their genotypes imputed to HD through the FImpute software. After quality control, remained for analysis 408,161 SNPs for IZ and 428.611 SNPs for Qualitas. The GWAS was performed for each population individually, considering the GBLUP methodology. Single-trait models were implemented in the analyzes, including, in addition to the random effects of animal and residual, the systematic effects of contemporary groups (CG), which were defined as: sex, year of birth and pen for the IZ, and year of test and pen for the Qualitas. For IZ, there were also considered, for all traits, the fixed effects of month of birth and, as covariable, age of animal (linear effect), age of dam (linear and quadratic effects) and the first two principal components (calculated based on the G matrix). For ADG and DMI, the quadratic effect of age of animal, as covariable, was added to the model. For Qualitas, it was also included in the model, for all traits, the linear effect of the animal age as covariable. In GWAS, some chromosomal regions of greater effect were found for each trait in both populations. However, no common regions were found. In GS, ten different approach and schemes involving the two Nellore populations were used to compare the accuracy of genomic prediction. In general, genomic predictions combining both populations are feasible, but, the benefits will depend on the trait and validation scheme. / CNPq: 132884/2016-0 / FAPESP: 2016/24228-9 / FAPESP: 2017/13411-0

Submitted by Reginaldo Soares de Freitas (reginaldo.freitas@ufv.br) on 2017-09-06T16:08:53Z No. of bitstreams: 1 texto completo.pdf: 542389 bytes, checksum: 16711f10ab30fbe3018e7a5d3be9aa19 (MD5) / Made available in DSpace on 2017-09-06T16:08:53Z (GMT). No. of bitstreams: 1 texto completo.pdf: 542389 bytes, checksum: 16711f10ab30fbe3018e7a5d3be9aa19 (MD5) Previous issue date: 2017-04-28 / Conselho Nacional de Desenvolvimento Científico e Tecnológico / A seleção de genitores é a primeira etapa em um programa de melhoramento e define o potencial de sucesso no desenvolvimento de uma cultivar superior. A seleção genômica ampla (do inglês, genome wide selection – GWS) associa informações moleculares e fenotípicas e prediz o desempenho de progênies futuras (valor genético genômico predito) com informações moleculares. Neste cenário, o objetivo deste estudo foi avaliar o potencial da GWS na predição do desempenho de híbridos e, consequentemente, na seleção de genitores para cruzamentos, utilizando-se somente informações genotípicas dos híbridos. Ademais, almejou-se comparar os resultados com técnicas tradicionais de seleção de genitores (seleção univariada e multivariada), visando o estabelecimento de blocos de cruzamento, para, então, parametrizar o potencial da GWS. Para execução do experimento foram utilizados dados simulados tomando como referência o genoma da soja. Foram criadas 200 RILs (do inglês, recombinant inbred line) com informações moleculares de 5400 SNPs e quatro características fenotípicas (produtividade, altura de planta, acamamento e ramificações laterais). Além disso, foram gerados 19900 híbridos oriundos do intercruzamento de todas as RILs. O conjunto de informações das RILs foram criadas fazendo alusão ao processo de fenotipagem e genotipagem de um conjunto de linhagens de soja com potencial para serem selecionadas como genitores em um programa de melhoramento. Primeiramente, foram selecionadas 10 linhagens seguindo critério univariado para a característica de interesse, ou seja, baseando-se no desempenho fenotípico per se para a produtividade. Em seguida, utilizou-se de um critério multivariado para seleção de genitores. Neste, foi empregado o método de agrupamento de Tocher, utilizando-se a distância euclidiana média, e a técnica de componentes principais para seleção de 10 linhagens geneticamente divergentes. Posteriormente, a seleção de genitores (10 linhagens) foi baseada no valor genético genômico estimado das RILs, cuja estimação foi obtida a partir da metodologia de seleção genômica ampla, considerando o carácter produtividade. Neste caso, foram selecionadas as linhagens com maior valor genético genômico estimado. Os três blocos de cruzamentos gerados foram avaliados pelo modelo de análise dialélica proposto por Griffing (1956). Paralelamente, os valores fenotípicos dos híbridos foram analisados visando a seleção das melhores populações híbridas de cada bloco de cruzamento. Além disto, estimou-se o valor genético genômico predito de todos os híbridos oriundos do intercruzamento das 200 linhagens e o potencial preditivo foi verificado analisando-se o desempenho fenotípico dos melhores híbridos preditos. As metodologias de seleção de genitores univariada com base em valores fenotípicos e predição pela GWS apresentaram quatro populações híbridas promissoras de acordo com a análise dialélica, enquanto que o critério multivariado para seleção de genitores rendeu seis híbridos superiores. Entretanto, considerando-se a seleção dos 20% melhores híbridos baseados apenas em dados fenotípicos, foi observado que as populações mais promissoras foram encontradas no bloco de cruzamento baseado no valor genético genômico estimado dos genitores, seguido pelo critério multivariado e univariado. O desempenho fenotípico médio destes híbridos superiores foi 1,14, 1,11 e 0,93, respectivamente. A predição de performance empregada pela GWS para quatro características fenotípicas apresentou resultados promissores. O desempenho fenotípico dos melhores híbridos preditos para as características produtividade, altura de planta, acamamento e ramificações laterais apontou que 30%, 47%, 46% e 46% dos melhores híbridos preditos apresentaram excelente desempenho fenotípico, respectivamente. Além disso, observou-se que os genitores das populações híbridas com excelente desempenho fenotípico apresentavam elevados valores genético genômicos, destacando a importância de se considerar informações de genitores. Este resultado ressalta o potencial da GWS na predição do desempenho de híbridos e, consequentemente, na determinação dos genitores selecionados para cruzamentos. / Parental selection is the main stage in a breeding program, once it delimits the success in developing a new cultivar. The Genome Wide Selection (GWS) enables the association of molecular information with phenotypic data and predicts the performance of future progenies (estimated breeding value) using molecular information. In this scenario, the aim of this study was to evaluate the potential of GWS to predict hybrid performance and, consequently, support parental selection, only employing genotypic information of hybrid populations. Besides, we aimed to compare the results with traditional methods of parental selection (univariate and multivariate selection), in order to form crossing blocks, and, therefore, to parametrize the GWS potential. We ran this study based on simulated data from the soybean reference genome. We created 200 RILs (recombinant inbred line) associated to molecular information (5400 SNP markers) and four phenotypic traits (yield, plant height, lodging and number of branches). We created the 19900 hybrids from the intercross of all RILs as well. The group of RILs data was created aiming to allude the process of genotyping and phenotyping a set of soybean inbred lines with potential to yield promising hybrids. First, 10 inbred lines were selected according to per se performance criteria for yield, that is, the most yielding lines were selected. Then, a multivariate approach was employed for parental selection. In this case, the Tocher grouping technique, based on average Euclidean distance, and Principal Components analysis were employed to select the most genetically divergent inbred lines (10 inbred selected). Next, the parental selection (10 inbred lines selected) was based on estimated breeding value of RILs, whose estimation was made according to the GWS methodology for yield. All three crossing blocks were evaluated through diallele cross analysis following the method proposed by Griffing (1956). Alongside, the hybrids phenotypic performance was analyzed solely as well. Moreover, we estimated all hybrids breeding value from the intercross of the 200 RILs and its prediction capability was verified analyzing the phenotypic performance of the best predicted hybrids. The parental selection approaches based on univariate criteria and GWS prediction displayed four promising hybrid populations according to the diallele cross analysis, while the multivariate criteria yielded six superior hybrids. However, considering selecting the 20% best hybrids based solely on phenotypic performance, we observed that the most promising ones were found in the crossing block based on estimated breeding value, follow by the multivariate approach and univariate criteria. The phenotypic average performance of these superior hybrids populations were 1.14, 1.11 and 0.93, respectively. The performance prediction employed for four agronomic traits by GWS delivered promising results. The phenotypic analysis of the best hybrids according to the GWS prediction model (ones with highest estimated breeding value) for yield, plant height, lodging and number of branches pointed out that 30%, 47%, 46% and 46% of theses hybrids performed phenotypically greatly, respectively. In addition, the genitors of the hybrid populations with excellent phenotypic performance had great estimated breeding value, indicating parental information importance. These results highlight the potential of GWS in predicting the best hybrids and, therefore, establishing the best parents for crossing.

Despite the success of vaccines in routine use worldwide, there are substantial challenges hampering our ability to develop vaccines against extant diseases including malaria and tuberculosis. Novel approaches are urgently required to help us understand immunological correlates of protection against disease and facilitate our understanding of the impact of human genetic variation on the success of diverse vaccines. To identify host genetic factors responsible for variation in antibody responses against vaccine antigens delivered routinely to infants worldwide I performed a genome-wide association study (GWAS) involving 2,499 infants recruited from three diverse sites across Africa. I identified strong genetic associations between variants in the class II major histocompatibility complex (MHC) locus and responses against five antigens: pertussis toxin (PT), filamentous haemagglutinin (FHA) and pertactin; diphtheria toxin (DT); and hepatitis B surface antigen. To characterise these associations at the gene and allelic level I developed a large, high-resolution (6-digit 'G') population-specific human leukocyte antigen (HLA) imputation reference panel including 697 individuals from the vaccine GWAS typed at 11 genes, highlighting the diversity of HLA across the African continent. Using this panel I imputed HLA into the remaining GWAS dataset to fine-map the associations to specific HLA alleles, amino acid and single nucleotide polymorphism sites; some of which were found to be African specific. I then used these HLA association findings observed with PT response to correlate, through genetics, this trait with susceptibility to whooping cough in an independently recruited and analysed set of cohorts from the UK. I further used these genetic correlations to demonstrate the relevance of levels of PT-specific circulating follicular helper T-cells and TRBV29-1 T-cell receptor gene expression levels in the development of this protective immune response against PT. By using HLA-peptide binding studies I also demonstrate the diversity of mechanisms that are involved in HLA-disease association, showing that the breadth and affinity of DT-peptide binding are increased with HLA-DRB1 alleles associated with increased DT antibody responses. Taken together, these data represent the first comprehensive genetic association study of multiple vaccine responses undertaken in African infants. These results highlight the importance of human genetics in modulating protective responses against vaccine antigens and demonstrate how such associations can be harnessed to understand biological mechanisms of protective efficacy in greater detail that may in turn facilitate future vaccine development.

Comme pour toute pathologie, la survenue d’un cancer du sein est conditionnée par l’association de facteurs génétiques héréditaires et de facteurs environnementaux acquis. Les facteurs génétiques connus comprennent à la fois des mutations pathogènes rares induisant un risque élevé de développer un cancer du sein et des variants génétiques fréquents (single nucleotides polymorphisms - SNP) responsables d’une faible augmentation du risque. L’ensemble des résultats de ce manuscrit plaide en faveur d’un impact majeur des facteurs génétiques constitutionnels à la fois en ce qui concerne le risque de développer un cancer du sein mais aussi en tant que déterminants du type de cancer du sein, voire du pronostic. La survenue d’un cancer du sein exprimant les récepteurs aux estrogènes et HER2-négatif est associée à 4 SNP introniques du gène FGFR2. Le pronostic des cancers du sein n’est pas associé aux variants impliquant un risque de développer un cancer. Quatre SNP indépendants sont associés à une évolution péjorative des cancers du sein triple-négatifs.La séquence d’événements qui mène du génome du patient à celui de la tumeur reste complexe, mal connue et probablement spécifique à chaque cancer comme l’illustrent les deux cas liés à des mutations germinales BRCA1/2 étudiés en deuxième partie de manuscrit. Le dernier travail permet de faire un lien vers la pratique clinique et rapporte une prévalence des mutations germinales BRCA1/2 d’environ 3% dans une cohorte prospective de patientes présentant un cancer du sein métastatique, non sélectionnées en fonction de leur âge, type de cancer ou antécédents familiaux. / As in any disease, the development of breast cancer depends on genetic hereditary factors and environmental acquired factors. Genetic factors of breast cancer involve rare pathogenic mutations with high risk of developing a breast cancer and frequent genetic variants (single nucleotides polymorphisms - SNP) responsible for a low increase in the risk of cancer. The works presented in this manuscript show that germline genetic factors strongly determine the risk of developing a breast cancer, but also the subtype of breast cancer and may impact the prognosis. Estrogen-positive, HER2-negative breast cancer development is associated with 4 intronic SNP in FGFR2 gene. Breast cancer prognosis is not associated with variants conferring a risk of developing a breast cancer. Four independent SNP are associated with bad outcomes in triple-negative breast cancers.The way that leads from patient genome to tumor genome is complex, mainly unknown and probably different for each case, as illustrated in the two case reports involving BRCA1/2 germline mutations described in the second part of the manuscript. Last work is a clinical research trial and shows a prevalence of BRCA1/2 mutations of around 3%, in a prospective cohort with metastatic breast cancer patients unselected on their age, cancer type or family history.