From a fresh data science perspective, this thesis discusses the prediction of coronary artery disease based on Single-Nucleotide Polymorphisms (SNPs) from the Ontario Heart Genomics Study (OHGS). First, the thesis explains the k-Nearest Neighbour (k-NN) and Random Forest learning algorithms, and includes a complete proof that k-NN is universally consistent in finite dimensional normed vector spaces. Second, the thesis introduces two dimensionality reduction techniques: Random Projections and a new method termed Mass Transportation Distance (MTD) Feature Selection. Then, this thesis compares the performance of Random Projections with k-NN against MTD Feature Selection and Random Forest for predicting artery disease. Results demonstrate that MTD Feature Selection with Random Forest is superior to Random Projections and k-NN. Random Forest is able to obtain an accuracy of 0.6660 and an area under the ROC curve of 0.8562 on the OHGS dataset, when 3335 SNPs are selected by MTD Feature Selection for classification. This area is considerably better than the previous high score of 0.608 obtained by Davies et al. in 2010 on the same dataset.

La santé cardiovasculaire, la fonction immunitaire, l'hémostase et la réponse à d'autres maladies dépendent de l'abondance et des caractéristiques spécifiques des cellules sanguines. Au fil des années, un effort considérable a été fait pour trouver les variants génétiques, les gènes et les mécanismes de régulation impliqués dans la création de ces cellules. L'inactivation d'un allèle, appelée "perte de fonction" (LoF), est un type de variant codant que nous aimerions associer aux phénotypes sanguins. Comme ces mutations ne peuvent pas être artificiellement induites chez l'humain, pour des raisons éthiques évidentes, nous observons les occurences naturelles de ces pertes de fonction et espérons que la taille des cohortes sera suffisante pour trouver des associations statistiquement significatives. L'inactivation des deux allèles, appelée "knockout" (KO), peut avoir des conséquences plus fortes qu'une simple perte de fonction. Nous espérons également trouver des KO d'origine naturelle grâce à la taille des cohortes. La combinaison de deux variants LoF différents sur les deux allèles est appelée knockout hétérozygote composé. Nous nous intéressons également aux variants non codants qui affectent l'expression des gènes impliqués dans l'hématopoïèse. Certains de ces variants créent ou perturbent des sites de liaison des facteurs de transcription (TF), ces protéines qui se lient à des séquences d'ADN spécifiques et régulent l'expression des gènes. Les sites de liaison (TFBS) des facteurs de transcription se trouvent dans les promoteurs des gènes et dans les amplificateurs spécifiques au type cellulaire. Alors que certaines de ces mutations peuvent être bénignes ou même bénéfiques, la présence d'un LoF ou d'un KO peut être trop nuisible à la survie de l'individu. Les résultats de cette étude sont limités par le biais de survie. Comparée à une étude d'association pangénomique, cette étude se concentre sur un plus petit nombre de variants génétiques pour augmenter la puissance statistique et offrir une interprétation pour les résultats statistiquement significatifs. Le programme Trans-Omics for Precision Medicine (TOPMed) recueille et garantit la qualité des 45 000 séquences du génome entier que nous avons utilisées dans cette étude, ainsi que les bilans sanguins correspondants. Grâce à ces données, nous avons pu trouver plusieurs associations connues et nouvelles entre des variants rares et des phénotypes sanguins. / Cardiovascular health, immune function, hemostasis and the response to other illnesses depend on the abundance and specific features of blood cells. Over the years, a considerable effort has been made to find which genetic variants, genes and regulatory mechanisms are involved in the creation of these cells. The inactivation of an allele, called a loss-of-function (LoF), is a type of coding variant we would like to associate with blood phenotypes. For obvious ethical reasons, these mutations cannot be artificially induced in human, so we fall back on natural occurrences and hope that large cohorts will provide enough samples to find statistically significant associations. The inactivation of both alleles, called a knockout (KO), may have stronger consequences than a simple loss-of-function. We also hope to find naturally occurring knockouts thanks to the size of a large cohort. The combination of two different LoF variants is called a compound heterozygote knockout. We are also interested in non-coding variants that affect the expression of genes that are involved in hematopoiesis. Some of these variants create or disrupt the binding sites of transcription factors (TF), the proteins that bind to specific DNA sequences and regulate gene expression. Transcription factors binding sites (TFBS) are found in gene promoters and cell type specific enhancers. While some of these mutations can be benign or even beneficial, the presence of a LoF or KO may be too detrimental for the individual to survive. The results of this study are limited by survival bias. Compared to a genome-wide association study, this study focuses on a smaller number of genetic variants to increase statistical power and give an interpretation to the statistically significant findings. The Trans-Omics for Precision Medicine (TOPMed) program collects and ensures the quality of the 45,000 whole-genome sequences we used in this study, as well as the corresponding complete blood counts. Thanks to this raw data, we were able to find several known and novel associations between rare variants and blood phenotypes.

<p>Within the last few decades, swine breeding programs have been refined to include pork quality and novel carcass traits alongside growth, feed efficiency, and carcass leanness in the selection programs for terminal sire lines with a goal to produce high quality and efficient pork product for consumers. In order to accurately select for multiple traits at once, it becomes imperative to explore their genetic and biological architecture. The genetic architecture of traits can be explored through the estimation of genetic parameters, genome-wide association studies (GWAS), gene networks and metabolic pathways. An alternative approach to explore the genetic and biological connection between traits is based on principal component analysis (PCA), which generates novel “pseudo-phenotypes” and biological types (biotypes). In this context, the main objective of this thesis was to understand the genetic and biological relationship between three growth, eight conventional carcass, 10 pork quality, and 18 novel carcass traits included in two studies. The phenotypic data set included 2,583 records from female Duroc pigs from a terminal sire line. The pedigree file contained 193,764 animals and the genotype file included 21,344 animals with 35,651 single nucleotide polymorphisms (SNPs). The results of the first study indicate that genetic progress can be achieved for all 39 traits. In general, the heritability estimates were moderate, while most genetic correlations were generally moderate to high and favorable. Some antagonisms were observed but those genetic correlations were low to moderate in nature. Thus, these relationships can be considered when developing selection indexes. The second study showed that there are strong links between traits through their principal components (PCs). The main PCs identified are linked to biotypes related to growth, muscle and fat deposition, pork color, and body composition. The PCs were also used as pseudo-phenotypes in the GWAS analysis, which identified important candidate genes and metabolic pathways linked to each biotype. All of this evidence links valuable variables such as belly, color, marbling, and leanness traits. Our findings greatly contribute to the optimization of genetic and genomic selection for the inclusion of valuable and novel traits to improve productive efficiency, novel carcass, and meat quality traits in terminal sire lines.<br></p><p></p>

L'un des objectifs centraux de la biologie est de comprendre comment l'ADN, la séquence primaire, donne lieu à des traits observables. À cette fin, nous examinons ici des méthodes pour identifier les composants génétiques qui influencent les traits microbiens. Par « identifier », nous entendons l'élucidation à la fois l'état allélique et de la position physique de chaque variante causale d'un phénotype d'intérêt à la résolution des nucléotides de paires de bases. Nous nous sommes concentrés sur les études d'association génomique (genome-wide association studies; GWAS) en tant qu'approche générale d’étudier l'architecture génétique des traits. L'objectif global de cette thèse était d'examiner de manière critique les méthodologies GWAS et de les considérer en pratique dans des populations microbiennes fortement clonales et non- clonales (i.e. avec recombinaison fréquent). Le domaine de la GWAS microbienne est relativement nouveau par rapport aux quinze dernières années de la GWAS humaine, et en tant que tel, nous avons commencé par un examen de l'état de la GWAS microbienne. Nous avons posé deux questions principales : 1) Les méthodes GWAS humaines fonctionnent-elles facilement et sans modification pour les populations microbiennes ? 2) Et sinon, quels sont les problèmes méthodologiques centraux et les modifications nécessaires pour la GWAS microbienne? À partir de ces résultats, nous avons ensuite détaillé le déséquilibre de liaison (linkage disequilibrium; LD) comme principal obstacle dans la GWAS microbien, et nous avons présenté une nouvelle méthode, POUTINE, pour relever ce défi en exploitant les mutations homoplasiques pour briser implicitement la structure LD. Le reste de la thèse présente à la fois les méthodes traditionnelles GWAS (comptage des allèles) et POUTINE (comptage d’homoplasies) appliquées à une population hautement recombinogène de génomes de vibrions marins. Malgré une taille d'échantillon modeste, nous donnons un premier aperçu de l'architecture génétique de la résistance aux bactériophages dans une population naturelle, tout en montrant que les récepteurs des bactériophages jouent un rôle primordial. Ce résultat est en pleine cohérence avec des expériences en laboratoire de coévolution phage-bactérie. Il est important de noter que cette architecture met en évidence à quel point la sélection positive peut sculpter certains traits microbiens différemment de nombreux traits complexes humains, qui sont généralement soumis à une faible sélection purificatrice. Plus précisément, nous avons identifié des mutations à effet important à haute fréquence qui sont rarement observées dans les phénotypes complexes humains où de nombreuses mutations à faible effet contribuent à l'héritabilité. La thèse se termine par des perspectives sur les voies à suivre pour la GWAS microbienne. / One of the central goals of biology is to understand how DNA, the primary sequence, gives rise to observable traits. To this aim, we herein examine methods to identify the genetic components that influence microbial traits. By "identify" we mean the elucidation of both the allelic state and physical position of each causal variant of a phenotype of interest down to the base-pair nucleotide resolution. Our focus has been on genome-wide association studies (GWAS) as a general approach to dissecting the genetic architecture of traits. The overarching aim of this thesis was to critically examine GWAS methodologies and to consider them in practice in both strongly clonal and highly recombining microbial populations. The field of microbial GWAS is relatively new compared to the over fifteen years of human GWAS, and as such, we began this work with an examination of the state of microbial GWAS. We asked and attempted to answer two main questions: 1) Do human GWAS methods readily work without modification for microbial populations? 2) And if not, what are the central methodological problems and changes that are required for a successful microbial GWAS? Building from these findings, we then detailed linkage disequilibrium (LD) as the primary obstacle in microbial GWAS, and we presented a new method, POUTINE, to address this challenge by harnessing homoplasic mutations to implicitly break LD structure. The remainder of the thesis showcases both traditional GWAS methods (allele counting) and POUTINE applied to a highly recombining population of marine vibrio genomes. Despite a small sample size, we provide a first glimpse into the genetic architecture of bacteriophage resistance in a natural population and show that bacteriophage receptors play a primary role consistent with experimental populations of phage-bacteria coevolution. Importantly, this architecture highlights how strong positive selection can sculpt some microbial traits differently than many human complex traits, which are generally under weak purifying selection. Specifically, we identified common frequency, large-effect mutations that are rarely observed in human complex phenotypes where many low-effect mutations are thought to contribute to the bulk of heritability. The thesis concludes with perspectives on ways forward for microbial GWAS.

No description available.

No description available.

No description available.

Cranberry common bean (Phaseolus vulgaris L.) is planted worldwide and consumed as a critical food source of human protein, fibre, carbohydrates, and minerals. Marsh spot (MS) is a physiogenic disorder which severely impacts seed quality in common beans. Previous studies indicate that MS involves a nutritional disorder caused by Mn deficiency. However, the inheritance and genetic mechanism of MS resistance are still not fully understood. To investigate the genetics of MS resistance, a population of 138 recombinant inbred lines (RILs) was developed from a bi-parental cross between a susceptible cultivar Messina and a resistant cultivar Cran09. The population and its two parents were evaluated for MS resistance during five consecutive years from 2015 to 2019 in both sandy and heavy clay soils in Morden, Manitoba, Canada. The severities of MS were rated and subsequently converted to MS resistance index (MSRI) and MS incidence (MSI). Statistical analyses indicated that MSI and MSRI were highly correlated (r = 0.96-0.99) and had high broad-sense heritability (H²) of 86.5% and 83.2%, respectively. Joint segregation analysis (JSA) of 18 phenotypic datasets from five years and two soil types showed that MS resistance was controlled by four major genes with genetic interactions - one of which may suppress the additive effect of the other three genes. To identify the quantitative trait loci (QTL) and the candidate genes associated with the MS resistance, the 138 RILs and the two parents were sequenced using genotyping by sequencing approach. A total of 52,676 SNPs were detected. After further filtering with a threshold of minor allele frequency > 0.01 and call rate > 20%, 2,061 SNPs were retained and then imputed for genetic map construction and QTL mapping. A genetic map consisting of 2,058 SNP markers on 11 linkage groups or chromosomes was constructed, which covered 1,004 recombination blocks with a total length of 6,449 cM and an average block of 6.42 cM. Three linkage map-based QTL-mapping models ICIM-ADD, ICIM-EPI, and GCIM and one genome-wide association study (GWAS) model RTM-GWAS for 18 phenotypic datasets from different years and soil types were used for identification of QTL. A total of 36 QTL, including 21 of additive and 15 of epistatic effects, were identified. Functional gene annotation analysis revealed 151 Mn-related candidate genes across the common bean reference genome and 17 of them harbored the six QTL discovered in this study. In conclusion, MS resistance in common bean is a highly heritable trait and controlled by several major and minor genes. The results of JSA and QTL mapping advance the current understanding of the genetic mechanisms of MS resistance in cranberry common bean, and provide additional resources for application in genomics-assisted breeding and potential isolation and functional characterization of the candidate genes.

Genome-wide association studies (GWASs) have reported numerous associations between risk variants and major psychiatric disorders (MPDs) including schizophrenia (SCZ), bipolar disorder (BPD), major depressive disorder (MDD) and others. We reviewed all of the published GWASs, and extracted the genome-wide significant (p<10) and replicated associations between risk SNPs and MPDs. We found the associations of 6 variants located in 6 genes, including L type voltage-gated calcium channel (LTCCs) subunit alpha1 C gene (), that were genome-wide significant ( ) and replicated at single-point level across at least two GWASs. Among them, the associations between MPDs and rs1006737 within are most robust. Thus, as a next step, the expression of the replicated risk genes in human hippocampus was analyzed. We found had significant mRNA expression in human hippocampus in two independent cohorts. Finally, we tried to elucidate the roles of venlafaxine and ω-3 PUFAs in the mRNA expression regulation of the replicated risk genes in hippocampus. We used cDNA chip-based microarray profiling to explore the transcriptome-wide mRNA expression regulation by ω-3 PUFAs (0.72/kg/d) and venlafaxine (0.25/kg/d) treatment in chronic mild stress (CMS) rats. ω-3 PUFAs and venlafaxine treatment elicited significant up-regulation. We concluded that might confer the genetic vulnerability to the shared depressive symptoms across MPDs and CACNA1C might be the therapeutic target for depressive endophenotype as well.

京都大学 / 新制・課程博士 / 博士(農学) / 甲第24654号 / 農博第2537号 / 新制||農||1097(附属図書館) / 学位論文||R5||N5435(農学部図書室) / 京都大学大学院農学研究科農学専攻 / (主査)教授 田尾 龍太郎, 教授 土井 元章, 准教授 中野 龍平 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DGAM