This dissertation aims at investigating the association between genotypes and phenotypes in human. Both common and rare regulatory variants have been studied. The phenotypes include disease risk, clinical traits and gene expression levels. This dissertation describes three different types of association study. The first study investigated the relationship between common variants and three sub-clinical traits as well as three complex diseases in the Center for Health Discovery and Well Being study (CHDWB). The second study is GWAS analysis of TNF-α and BMI/CRP conducted as a contribution to meta-GWAS analyses of these traits with investigators at the University of Groningen in the Netherlands, and the 1000 Genomes Consortium. The third study was the most original contribution of my thesis as it assessed the association between rare regulatory variants in promoter regions and gene expression levels. The results clearly show an enrichment of rare variants at both extremes of gene expression. This dissertation provides insight into how common and rare variants associate with broadly-defined quantitative phenotypes. The demonstration that rare regulatory variants make a substantial contribution to gene expression variation has important implications for personalized medicine as it implies that de novo and other rare alleles need to be considered as candidate effectors of rare disease risk.

As fissuras orofaciais não-sindrômicas (FO-NS) correspondem a 70% de todos os casos de FO, possuem etiologia complexa e pouco compreendida, sendo consideradas de herança multifatorial com forte influência de fatores genéticos e ambientais. Apesar de estudos de análise de ligação e associação apontarem vários loci de susceptibilidade às FO-NS, o componente genético ainda não está totalmente explicado. Fatores ambientais também possuem um importante papel na etiologia das FO, e alguns já foram replicados em várias populações. Fatores como exposição materna ao álcool, drogas, tabaco, medicamentos, desnutrição e baixo nível socioeconômico são alguns dos fatores já associados a esta condição. As infecções periodontais são comuns em mulheres grávidas e estão associadas a parto prematuro, baixo peso fetal e, mais recentemente, foram reportadas como fator de risco aumentado para FO-NS nos fetos. Adicionalmente, o avanço das tecnologias de sequenciamento do DNA melhorou exponencialmente a compreensão do microbioma humano e sua influência no estado de saúde e doença, e, mais especificamente, o conhecimento sobre o impacto do microbioma na gravidez. O objetivo deste projeto foi identificar novos fatores etiológicos genéticos e ambientais das FO-NS. Para isso, primeiramente, sequenciamos 68 genes candidatos a FO por sequenciamento de nova geração em 193 indivíduos com FO-NS familial. Nós encontramos enriquecimento significativo de variantes raras e patogênicas de perda de função nos indivíduos com FO-NS e observamos que essas variantes estão em genes intolerantes a esse tipo de mutação. Também reportamos novas variantes raras do tipo perda de função no gene ARHGAP29 e sua importância na susceptibilidade as FO-NS familiais. Além disso, sugerimos o uso de um ponto de corte baseado no escore pLI do banco de dados ExAC como parâmetro para priorizar variantes em estudos de FO-NS familiares, assumindo modelo de herança mono ou oligogênico. Adicionalmente, estudamos o microbioma oral de mães de crianças com FO-NS e mães de crianças sem malformações, utilizando o sequenciamento da subunidade 16S do rRNA das bactérias com o objetivo de verificar diferenças consistentes na composição do microbioma oral de mães de crianças com FO-NS, levando em consideração a presença ou não de doenças infecciosas periodontais maternas. A casuística foi composta de 6 mães de recém-nascidos de até 1 mês que apresentaram FO-NS ao nascimento e mães de crianças sem qualquer malformação congênita. As análises de alfa e beta diversidades não demonstraram diferença significativa na composição do microbioma oral de mães de crianças com FO-NS e mães de crianças controle, contudo observamos que o grupo com infecções periodontais possui a diversidade taxonômica mais abundante do que o grupo hígido. Em resumo, nesse estudo piloto não foi possível identificar alterações no microbioma oral como um fator etiológico das FO-NS. Novas análises em uma casuística maior são necessárias para a confirmação desse achado / The non-syndromic orofacial clefts (nsOFC) correspond to 70% of all OFC cases, have complex etiology and are poorly understood, being considered multifactorial inheritance with a strong influence of genetic and environmental factors. Although linkage and association analysis studies point to several nsOFC susceptibility loci, the genetic component is not yet fully explained. Environmental factors also play an important role in OFC etiology, and some have been replicated in several populations. Factors such as maternal exposure to alcohol, drugs, tobacco, drugs, malnutrition and low socioeconomic status are some of the factors already associated with this condition. Periodontal infections are common in pregnant women and are associated with preterm birth, low birth weight and, more recently, have been reported as an increased risk factor for nsOFC in fetuses. Additionally, the advancement of DNA sequencing technologies has exponentially improved the understanding of the human microbiome and its influence on health and disease status, and, more specifically, knowledge about the impact of the microbiome on pregnancy. The objective of this project was to identify new genetic and environmental etiological factors of nsOFC. For this, we first sequenced 68 candidate genes by next generation sequencing in 193 individuals with familial nsOFC. We found significant enrichment of rare and pathogenic loss of function variants in individuals with nsOFC and we observed that these variants were in genes intolerant to this type of mutation. We also reported new rare loss-of-function variants in the ARHGAP29 gene and its importance in the liability of familial nsOFC. In addition, we suggested the use of a cutoff point based on the ExAC database pLI score as a parameter to prioritize variants in familial nsOFC studies, assuming a mono or oligogenic inheritance model. In addition, we studied the oral microbiome of 6 mothers of newborns up to 1-month-old with nsOFC and 6 mothers of newborns without congenital malformations using the 16S rRNA sequencing in order to verify consistent differences in the composition of the oral microbiome of mothers of children with nsOFC, taking into account the presence or absence of maternal periodontal infectious diseases. The analysis of alpha and beta diversities did not show a significant difference in the composition of the oral microbiome of mothers of nsOFC children and mothers of control children, however, we observed that the group with periodontal infectious diseases has more abundant taxonomic diversity than the healthy group. In summary, in this pilot study, it was not possible to identify alterations in the oral microbiome as an etiological factor of FO-NS. New analyzes in a larger cohort are necessary to confirm this finding

There is solid evidence that complex human diseases can be caused by rare variants. Next generation sequencing technology has revolutionized the study of complex human diseases, and made possible detecting associations with rare variants. Traditional statistical methods can be inefficient for analyzing sequence data and underpowered. In addition, due to high cost of sequencing, it is also necessary to explore novel cost effective studies in order to maximize power and reduce sequencing cost. In this thesis, three important problems for analyzing sequence data and detecting associations with rare variants are presented. In the first chapter, we presented a new method for detecting rare variants/binary trait associations in the presence of gene interactions. In the second chapter, we explored cost effective study designs for replicating sequence based association studies, combining both sequencing and customized genotyping. In the third chapter, we present a method for analyzing multiple phenotypes in selected samples, such that phenotypes that are commonly measured in different studies can be jointly analyzed to improve power. The methods and study designs presented are important for dissecting complex trait etiologies using sequence data.

<p>Although much research has focused on cognitive ability and the genetic and environmental factors that might influence it, this aspect of human nature is still far from being well understood. It has been well-established that certain factors such as age and education have significant impacts on performance on most cognitive tests, but the effects of variables such as cognitive pastimes and strategies used during testing have generally not been assessed. Additionally, no genetic variant has yet been unequivocally shown to influence the normal variation in cognitive ability of healthy individuals. Candidate gene studies of cognition have produced conflicting results that have not been replicable, and genome-wide association studies have not found common variants with large influences on this trait.</p><p>Here, we have recruited a large cohort of healthy volunteers (n=1,887) and administered a brief cognitive battery utilizing diverse, common, and well-known tests. In addition to providing standard demographic information, the subjects also filled out a questionnaire that was designed to assess novel factors such as whether they had seen the test before, in what cognitive pastimes they participated, and what strategies they had used during testing. Linear regression models were built to assess the effects of these variables on the test scores. I found that the addition of novel covariates to standard ones increased the percent of the variation in test score that was explained for all tests; for some tests, the increase was as high as 70%.</p><p>Next, I examined the effects of genetic variants on test scores. I first performed a genome-wide association study using the Illumina HumanHap 550 and 610 chips. These chips are designed to directly genotype or tag the vast majority of the common variants in the genome. Despite having 80% power to detect a common variant explaining at least 3-6% (depending on the test) of the variation in the trait, I did not find any genetic variants that were significantly associated after correction for multiple testing. This is in line with the general findings from GWA studies that single common variants have a limited impact on complex traits.</p><p>Because of the recent technological advances in next-generation sequencing and the apparently limited role of very common variants, many human geneticists are making a transition from genome-wide association study to whole-genome and whole-exome sequencing, which allow for the identification of rarer variants. Because these methods are currently costly, it is important to utilize study designs that have the best chance of finding causal variants in a small sample size. One such method is the extreme-trait design, where individuals from one or both ends of a trait distribution are sequenced and variants that are enriched in the group(s) are identified. Here, I have sequenced the exomes of 20 young individuals of European ethnicity: 10 that performed at the top of the distribution for the cognitive battery and 10 that performed at the bottom. I identified rare genetic variants that were enriched in one extreme group as compared to the other and performed follow-up genotyping of the best candidate variant that emerged from this analysis. Unfortunately, this variant was not found to be associated in a larger sample of individuals. This pilot study indicates that a larger sample size will be needed to identify variants enriched in cognition extremes.</p><p>Finally, I assessed the effect of topiramate, an antiepileptic drug that causes marked side effects in certain cognitive areas in certain individuals, on some of the healthy volunteers (n=158) by giving them a 100 mg dose and then administering the cognitive test two hours later. I compared their scores at this testing session to those at the previous session and calculated the overall level to which they were affected by topiramate. I found that the topiramate blood levels, which were highly dependent on weight and the time from dosing to testing, varied widely between individuals after this acute dose, and that this variation explained 35% of the variability in topiramate response. A genome-wide association study of the remaining variability in topiramate response did not identify a genome-wide significant association.</p><p>In sum, I studied the contributions of both environmental and genetic variables to cognitive ability and cognitive response to topiramate. I found that I could identify environmental variables explaining large proportions of the variation in these traits, but that I could not identify genetic variants that influenced the traits. My analysis of genetic variants was for the most part restricted to the very common ones found on genotyping chips, and this and other studies have generally found that single common genetic variants do not have large affects on complex traits. As we move forward into studies that involve the sequencing of whole exomes and genomes, genetic variants with large effects on these complex traits may finally be found.</p> / Dissertation

Rare disease variants are receiving increasing importance in the past few years as the potential cause for many complex diseases, after the common disease variants failed to explain a large part of the missing heritability. With the advancement in sequencing techniques as well as computational capabilities, statistical methodology for analyzing rare variants is now a hot topic, especially in case-control association studies. In this thesis, we initially present two related statistical methodologies designed for case-control studies to predict the number of common and rare variants in a particular genomic region underlying the complex disease. Genome-wide association studies are nowadays routinely performed to identify a few putative marker loci or a candidate region for further analysis. These methods are designed to work with SNP data on such a genomic region highlighted by GWAS studies for potential disease variants. The fundamental idea is to use Bayesian methodology to obtain bivariate posterior distributions on counts of common and rare variants. While the first method uses randomly generated (minimal) ancestral recombination graphs, the second method uses ensemble clustering method to explore the space of genealogical trees that represent the inherent structure in the test subjects. In contrast to the aforesaid methods which work with SNP data, the third chapter deals with next-generation sequencing data to detect the presence of rare variants in a genomic region. We present a non-parametric statistical methodology for rare variant association testing, using the well-known Kolmogorov-Smirnov framework adapted for genetic data. it is a fast, model-free robust statistic, designed for situations where both deleterious and protective variants are present. It is also unique in utilizing the variant locations in the test statistic.

La Polyarthrite Rhumatoïde (PR) est une maladie auto-immune inflammatoire complexe qui touche près de 0,3% de la population française. A ce jour, malgré l'identification d'un facteur génétique majeur (HLA-DRB1), et d'une centaine de facteurs de susceptibilité d'effet faible à modéré (majoritairement identifiés par des études d'associations pangénomiques - GWAS), on ne peut expliquer au plus que 50% de la composante génétique de la maladie. Les GWAS se focalisant sur les variants fréquents (fréquence de l'allèle mineure (MAF) ≥ 1%) et considérant l'effet de ces variants comme indépendants, nous avons cherché dans ces travaux à identifier de nouveaux facteurs génétiques de la PR via l'analyse de variants rares (variants d'un seul nucléotide (SNVs) ou petites insertions et délétions (InDels)) pour lesquels peu d'études sont référencées dans la littérature. Nous avons ensuite étudié les interactions gène/gène (GxG) dans des voies biologiques enrichies par les variants rares identifiés.Pour cela, nous avons travaillé sur deux jeux de données obtenus par séquençage d'exomes. Dans le premier échantillon (data1), nous avons cherché à évaluer la contribution des variants rares dans 1080 gènes candidats séquencés chez 240 cas et 240 témoins français. Dans le second échantillon (data2), notre objectif était d'identifier de nouveaux facteurs génétiques par l'analyse de variants rares chez 30 individus (dont 19 atteints) appartenant à 9 familles françaises à cas multiples de PR. Nous avons mis en place un workflow afin de réaliser toutes les étapes de traitement des séquences obtenues jusqu'à l'identification des variants (alignement des lectures sur la séquence de référence du génome humain GRCh37, nettoyage de l'alignement, identification des variants (SNV et Indels) et filtre qualité des variants identifiés).Avec data1, nous avons répliqué l'association entre la PR et le gène BTNL2 (p-value = 3,0E-6) et identifié trois nouveaux gènes à risque (p-value ≤ 4,0E-3), impliqués dans la différentiation et l'activation des cellules du système immunitaire, en combinant les multiples variants de fréquences faibles à modérées identifiés dans ces gènes (analyses d'association de type burden). Dans data2, nous avons effectué une étude d'association-liaison, par laquelle nous avons identifié 3 gènes - SUSD5, MNS1 et SMYD5 - présentant une agrégation de variants (rares et fréquents) associée à la PR (p-value < 0,04 après 10E6 permutations), et un gène nommé SUPT20H dont le signal d'association était porté par un seul variant rare à pénétrance complète dans une famille et sans phénocopie. Nous avons aussi mis en évidence, via une étude d'enrichissement, une agrégation de variants rares dans plusieurs voies biologiques dont l'adhésion focale. Dans cette voie, nous avons identifié 9 interactions candidates pour lesquelles plusieurs combinaisons génotypiques semblent conférer un risque supplémentaire de développer la PR (p-value ≤ 5,0E-5). / Rheumatoid arthritis (RA) is a complex inflammatory autoimmune disease affecting about 0.3% of French population. Today, despite the identification of a major genetic factor (HLA-DRB1), and more than one hundred susceptibility factors with low to moderate effect (mainly identified by Genome-Wide association studies - GWAS), we cannot explain more than 50% of RA genetic component. Knowing that GWAS only study frequent variants (minor allele frequency (MAF) ≥ 1%) and consider that all of them are independent, we tried to identify new RA genetic factors by focusing on rare variants (single nucleotide variants (SNVs) or small insertions and deletions (InDels)) for which, to date, only few studies has been conducted. In addition, we studied gene/gene interactions (GxG) in biological pathways enriched for rare susceptible variants.To this end, we worked on two datasets obtained by exome sequencing. With the first dataset (data1), we wanted to evaluate the contribution of rare variants to RA risk into 1080 candidate genes sequenced in 240 cases et 240 controls from French population. With the second dataset (data2), our aim was to identify new genetic factors by focusing on rare variants selected from 30 individuals (including 19 affected) belonging to 9 French multiplex families. We set up in the laboratory a workflow to process the produced sequences up to the identification of variants (read alignment on human reference genome GRCh37, alignment refinement, variant identification (SNV et Indels) and quality filters).In data1, we replicated the association between RA and BTNL2 gene (p-value = 3,0E-6) and identified 3 new RA risk genes (p-value ≤ 4,0E-3), involved in the differentiation and activation of immune system cells, by combining rare to low frequency variants (burden association analysis). In data2, with a linkage – association study, we identified 3 genes - SUSD5, MNS1 and SMYD5 – presenting an aggregation of rare and frequent variants associated with RA (p-value < 0.04 with 10E6 permutations), and another gene SUPT20H in which we identified one rare variant with complete penetrance in one of the family and without phenocopy. Finally, we identified, by enrichment analysis, several biological pathways presenting an aggregation of rare variants. In one of them (focal adhesion), we extracted 9 candidate GxG interactions for which multiple genotype combinations seem to increase RA risk (p-value ≤ 5,0E-5).

No description available.

Background: Type 2 Diabetes (T2D) is growing in prevalence worldwide over the last century. T2D incidence is linked to numerous complications, increased risk of heart disease, and oncology outcomes. This highlights the importance of preventive measures for T2D, wherein genetic predisposition can serve as an early warning sign. The role of rare variants (RVs) in T2D pathogenesis has not been adequately explored due to study size limitations, therefore we hypothesized that new associations could be found using publicly available data repositories. Methods: Significant RV gene burden for T2D risk was discovered using exome sequences obtained from the United Kingdom Biobank (UKB) (n=162,215), then tested for replication in the Korean Association Resource project (n=973), the Metabolic Syndrome in Men Study (n=969), the San Antonio Mexican American Family Studies (n=309), and a pooled meta-analysis of the latter three cohorts. RV gene burden was reassessed in secondary analyses using T2D cases from each cohort and summary level data from the Genome Aggregation Database (GnomAD) (n=125,748). Results: UKB exome wide significant associations were found in GCK (OR=2.44, p=8.91×10-11) and PAM (OR=1.32, p=1.39×10-6), and suggestive associations (p<0.001) were found in 33 additional genes. Replication was limited in KARE, METSIM, SAMAFS and in the secondary analyses with GnomAD because of limited sample sizes and miscalibration with the external control, respectively. Follow-up analyses include exploration of RV gene burden in additional diabetes subtypes, evaluation of clinical features between RV carriers and non-carriers, comparing the ability to predict T2D with rare variant, polygenic, and phenotypic risk scores. Methodological improvements include the incorporation of robust analytic tools and increasing access to a greater diversity and number of samples. Conclusion: Publicly available exome sequencing data has identified genes where RV burden affects T2D pathogenesis and risk. The study of rare genetic variation in diabetes is just beginning. / Thesis / Master of Science (MSc)

In the last years, Genome-Wide Associations Studies (GWAS) found many variants associated with complex diseases. However, the biological and molecular links between these variants and phenotypes are still mostly unknown. Also, even if sample sizes are constantly increasing, the associated variants do not explain all the heritability estimated for many traits. Many hypotheses have been proposed to explain the problem: from variant-variant interactions, the effect of rare and ultra-rare coding variants and also technical biases related to sequencing or statistic on sexual chromosomes. In this thesis, we mainly explore the hypothesis of variant-variant interaction and, briefly, the rare coding variants hypothesis while also considering possible molecular effects like allele-specific expression and the effects of variants on protein interfaces. Some parts of the thesis are also devoted to explore the implementation of efficient computational tools to explore these effects and to perform scalable genotyping of germline single nucleotide polymorphisms (SNPs) in huge datasets. The main part of the thesis regards the development of a new resource to identify putative variant-variant interactions. In particular, we integrated ChIP-seq data from ENCODE, transcription factor binding motifs from several resources and genotype and transcript level data from GTeX and TCGA. This new dataset allows us to formalize new models, to make hypothesis and to find putative novel associations and interactions between (mainly non-coding) germline variants and phenotypes, like cancer-specific phenotypes. In particular, we focused on the characterization of breast cancer and Alzheimer’s Disease GWAS risk variants, looking for putative variants’ interactions. Recently, the study of rare variants has become feasible thanks to the biobanks that made available genotypes and clinical data of thousands of patients. We characterize and explore the possible effects of rare coding inherited polymorphisms on protein interfaces in the UKBioBank trying to understand if the change in structure of protein can be one of the causes of complex diseases. Another part of the thesis explores variants as causal molecular effect for allele-specific expression. In particular, we describe UTRs variants that can alter the post-transcriptional regulation in mRNA leading to a phenomenon where an allele is more expressed than the other. Finally, we show those variants can have prognostic significance in breast cancer. This thesis work introduces results and computational tools that can be useful to a broad community of researcher studying human polymorphisms effects.

O transtorno do espectro do autismo (TEA) constitui um grupo heterogêneo e altamente prevalente de doenças do neurodesenvolvimento. Análises genômicas recentes têm revelado um grande número de variantes genéticas potencialmente deletérias nos pacientes com TEA, a maioria rara ou privada. Um enorme desafio atual é determinar quais dentre essas variantes são as que de fato estão envolvidas na etiologia do transtorno nos pacientes, e quantas variantes patogênicas são necessárias para a penetrância completa do TEA em cada paciente. Recentemente, por meio do sequenciamento completo do exoma de um subgrupo de pacientes com TEA não-sindrômico - nos quais observamos hiperfuncionamento da via de sinalização intracelular mTORC1 - identificamos que um dos pacientes (referido como F2688) é heterozigoto composto para variantes de substituição de aminoácidos raras e potencialmente deletérias no gene ELN. Este gene codifica Relina, uma grande glicoproteína de matriz extracelular que, por meio da ativação da proteína Dab1 e de diferentes vias de sinalização intracelular, controla a migração e o posicionamento dos neurônios, a arborização de neuritos, e o funcionamento das sinapses em várias regiões do encéfalo, tanto no desenvolvimento embrionário como na vida adulta. Estudos anteriores já haviam descrito variantes em heterozigose potencialmente de perda de função no gene RELN em pacientes com TEA; contudo, nenhum desses estudos investigou disfunção da sinalização Relina-Dab1 nos pacientes e, portanto, os efeitos moleculares e celulares de tais variantes sobre células neurais humanas ainda são poucos explorados. Neste trabalho, utilizando células neuroprogenitoras (NPCs) derivadas de células-tronco pluripotentes induzidas do paciente F2688, de outros pacientes com TEA sem mutação em RELN (n=5) e de indivíduos controles (n=5), nós descrevemos que as NPCs do paciente F2688 apresentam: i) disfunção da via de sinalização Relina-Dab1; ii) hiperfuncionamento da via de sinalização mTORC1; iii) crosstalk anormal entre as vias de sinalização Relina-Dab1 e mTORC1, o qual é atenuado com o uso da rapamicina, um inibidor específico de mTORC1. Portanto, nossos resultados sugerem, pela primeira vez, uma relação anormal entre as vias de sinalização Relina-Dab1 e mTORC1 em TEA não-sindrômico / Autism Spectrum Disorder (ASD) is a heterogeneous and highly prevalent group of neurodevelopmental disorders. Whole-genome-based approaches have generated catalogues of thousands of rare and potentially deleterious genetic variants in ASD patients. However, the challenge now is to identify genuine disease-causing/risk variants among the multitude of variants discovered in each exome/genome and how many variants are required to cause the disease. Recently, we performed whole-exome sequencing in a subgroup of ASD patients - in whom we found mTORC1 signaling hyperfunction - and identified rare and potentially deleterious compound heterozygous variants in the RELN gene in one patient (called as F2688). The RELN gene encodes Reelin, a large secreted glycoprotein that controls neuronal migration, layer formation, neurite outgrowth, and plasticity of synapses in both the developing and the adult brain. Evidence from previous studies suggests that certain potential loss-of-function variants in RELN gene can contribute to ASD susceptibility; however, few studies today have directly demonstrated impairment of the Reelin signal transduction cascade in ASD patients and therefore, the molecular and cellular effects of these variants in human neuronal cells are still poorly explored. Here, by using induced pluripotent stem cells derived neuronal progenitor cells from F2688 patient, from other ASD patients who do not carry RELN disrupting variants (n=5) and from control individuals (n=5), we have demonstrated that F2688-derived NPCs show: i) impaired Reelin-Dab1 signaling; ii) overactive mTORC1 signaling; iii) and abnormal crosstalk between mTORC1 and Reelin-Dab1 signaling pathways, which it attenuated by rapamycin (a specific mTORC1 inhibitor). Taken together, our results point to an abnormal interplay between mTORC1 and Reelin-Dab1 networks in nonsyndromic ASD