Master of Science / Department of Statistics / Haiyan Wang / Motivation: Genome-wide association studies (GWAS) provide an important avenue for investigating many common genetic variants in different individuals to see if any variant is associated with a trait. GWAS is a great tool to identify genetic factors that influence health and disease. However, the high dimensionality of the gene expression dataset makes GWAS challenging. Although a lot of promising machine learning methods, such as Support Vector Machine (SVM), have been investigated in GWAS, the question of how to improve the accuracy of the result has drawn increased attention of many researchers A lot of the studies did not apply feature selection to select a parsimonious set of relevant genes. For those that performed gene selections, they often failed to consider the possible interactions among genes. Here we modify a gene selection algorithm BMSF originally developed by Zhang et al. (2012) for improving the accuracy of cancer classification with binary responses. A continuous response version of BMSF algorithm is provided in this report so that it can be applied to perform gene selection for continuous gene expression dataset. The algorithm dramatically reduces the dimension of the gene markers under concern, thus increases the efficiency and accuracy of GWAS. Results: We applied the continuous response version of BMSF on the wheat phenotypes dataset to predict two quantitative traits based on the genotype marker data. This wheat dataset was previously studied in Long et al. (2009) for the same purpose but used only direct application of SVM regression methods. By applying our gene selection method, we filtered out a large portion of genes which are less relevant and achieved a better prediction result for the test data by building SVM regression model using only selected genes on the training data. We also applied our algorithm on simulated datasets which was generated following the setting of an example in Fan et al. (2011). The continuous response version of BMSF showed good ability to identify active variables hidden among high dimensional irrelevant variables. In comparison to the smoothing based methods in Fan et al. (2011), our method has the advantage of no ambiguity due to difference choices of the smoothing parameter.

Fusarium head blight (FHB) is a major challenge in global wheat production. In the United States, the disease is predominantly caused by the fungus Fusarium graminearum. Utilization of FHB-resistant wheat cultivars integrated with other measures such as fungicide application is the most effective approach for the management of this disease. This study aimed to 1) identify novel quantitative trait loci (QTL) for resistance to FHB in a Brazilian spring wheat cultivar ‘Surpresa’ through bi-parental mapping, 2) detect QTL for FHB resistance in a global panel of 233 spring wheat accessions by genome-wide association analysis (GWAS), and 3) localize genomic regions governing traits associated with virulence in Fusarium graminearum. Using phenotypic and genotypic data from 187 recombinant inbred lines derived from the cross between Surpresa and a susceptible spring wheat cultivar ‘Wheaton’, four QTL (Qfhb.ndwp-2AS, Qfhb.ndwp-2AL, Qfhb.ndwp-3B, and Qfhb.ndwp-4D) were mapped on chromosomes 2A, 3B, and 4D of Surpresa, respectively. Qfhb.ndwp-2AS, Qfhb.ndwp-2AL, and Qfhb.ndwp-3B were found to be novel based on physical locations of the markers tightly linked to these QTL. Two significant marker-trait associations (Qfhb.ndwp-3A and Qfhb.ndwp-2BL) were detected by GWAS of 233 spring wheat accessions, which conferred type II and type III FHB resistance and mapped on chromosomes 3A and 2B, respectively. Both QTL were novel based on the physical locations of tightly linked markers. GWAS of virulence and fungicide sensitivity using 183 F. graminearum isolates collected from North Dakota identified two significant marker-trait associations in chromosomes 1 and 3 for virulence, and two for fungicide sensitivity. The genes associated with virulence that were detected in this study were not previously reported. Identification of these novel genes in metabolic pathways of F. graminearum could help to develop new strategies for the management FHB.

Le virus de la panachure jaune (RYMV) est une contrainte majeure pour la riziculture en Afrique. Deux gènes contrôlant des résistances récessives ont précédemment été décrits : RYMV1, qui code pour eIF(iso)4G1, un facteur d’initiation de la traduction et RYMV2, qui code pour CPR5-1, un probable composant du pore nucléaire impliqué dans la régulation des mécanismes de défense. Cependant, la capacité du virus à contourner ces résistances justifie la caractérisation de sources de résistance originales, présentes dans les espèces de riz africain O. glaberrima et O. barthii. Trois approches complémentaires ont été mises en œuvre afin d’identifier les facteurs génétiques contrôlant ces résistances.Une approche de cartographie génétique dans des populations bi-parentales a permis l’identification du gène RYMV3, contrôlant la résistance de l’accession Tog5307 et sans doute également de l’accession Tog5672. Il s’agit de la première résistance dominante identifiée dans le pathosystème riz/RYMV. RYMV3 a été cartographié dans un intervalle de 15 kb où deux gènes sont annotés, dont un gène NB-LRR. Des comparaisons de séquences entre accessions résistantes et accessions sensibles suggèrent que le polymorphisme responsable de la résistance est une mutation ponctuelle dans le domaine LRR du gène NB-LRR.Les deux autres approches ont reposé sur l’exploitation de données de séquençage Illumina de 163 accessions O. glaberrima et 84 accessions O. barthii. Les accessions O. glaberrima ont été phénotypées à la fois pour la résistance élevée et pour la résistance partielle au RYMV, et une partie des accessions O. barthii a été évaluée pour la résistance élevée. L’analyse de la variabilité allélique aux trois gènes majeurs de résistance a permis l’identification d’un probable nouvel allèle de résistance à RYMV1 et de six à RYMV2. Ces allèles sont actuellement en cours de validation. D’autre part, une approche de génétique d’association réalisée sur 125 accessions O. glaberrima a mis en évidence deux QTL de résistance partielle sur les chromosomes 6 et 11, dont l’un colocalise, en première approche, avec le gène RYMV3.Ce travail a ainsi permis l’identification d’un gène majeur, de deux QTL et de nouveaux allèles de résistance qui contribuent à une meilleure compréhension des interactions riz/RYMV et sont utilisables en sélection pour améliorer la durabilité des variétés résistantes. / The Rice yellow mottle virus (RYMV) is a major constraint for rice production in Africa. Two genes controlling recessive resistances have been previously described : RYMV1 coding for eIF(iso)4G1, a translation initiation factor, and RYMV2, coding for CPR5-1, a probable component of the nuclear pore complex involved in the regulation of defense mechanisms. However, the virus' ability to overcome these resistances highlight the need to characterize new sources of resistance in the African rice species O. glaberrima and O. barthii. Three complementary approaches were carried out in order to identify the genetic factors controlling these resistances.A genetic mapping strategy in bi-parental populations led to the identification of the RYMV3 gene, controlling resistance in the Tog5307 accession and probably also in the Tog5672 accession. It is the first dominant resistance identified in the rice/RYMV pathosystem. RYMV3 mapped in a 15 kb interval in which two genes annotated occur, including one NB-LRR gene.The two other strategies used were based on the utilization of Illumina sequencing data of 163 O. glaberrima accessions and 84 O. barthii accessions. O. glaberrima accessions were phenotyped for both high and partial resistance to RYMV, and the high resistance of a portion of the O. barthii accessions was assessed. Analysis of allelic variability at the previously identified genes led to the identification of a probable new resistance allele at RYMV1 and of six others at RYMV2. These alleles are currently undergoing validation. Furthermore, a genome wide association study was carried out on 125 O. glaberrima accessions, revealing two partial resistance QTLs on chromosomes 6 and 11, including one colocalized with RYMV3.This work has thus allowed the identification of one major resistance gene, of two QTLs and of new resistance alleles, contributing to a better understanding of rice/RYMV interactions and creating new prospects for the breeding for resistant varieties.

Understanding complex traits is a fundamental challenge in plant genetics and a prerequisite for molecular breeding. Tools for trait dissection are markers, maps, and quantitative trait locus (QTL) analysis. Marker-assisted selection (MAS) is an application that integrates these tools. In this thesis research, a new sequence-based marker was evaluated, maps were constructed and used, and QTLs were detected using two types of populations. Marker-assisted selection was used to develop a novel class of barley. Restriction-site Associated DNA (RAD), a sequence based-marker technology, allows for simultaneous high-density single nucleotide polymorphism (SNP) discovery and genotyping. We assessed the value of RAD markers for linkage map construction using the Oregon Wolfe Barley (OWB) mapping population. We compared a RAD-based map to a map generated using Illumina GoldenGate Assay (EST-based SNPs). The RAD markers generated a high quality map with complete genome coverage. We then used the RAD map to locate QTL for agronomic fitness traits. A paper describing this research was published (Chutimanitsakun et al., 2011). Marker-assisted selection was used to rapidly develop fall-sown barley germplasm for human food uses. The target traits were high grain β-glucan, vernalization sensitivity (VS) and low temperature tolerance (LTT). The target loci were WX and VRN-H2. Marker-assisted selection was effective in fixing target alleles at both loci and waxy starch led to increase in grain β-glucan. Unexpected segregation at VRN-H1 and VRN-H3, revealed by genome-wide association mapping (GW-AM), led to unanticipated phenotypic variation in VS and LTT. We found that GW-AM is an efficient and powerful method for identifying the genome coordinates of genes determining target traits. Precise information is obtained with perfect markers; additional research may be needed when multiple alleles are segregating at target loci and significant associations are with markers in linkage disequilibrium (LD) with the target loci. A paper describing this research will be submitted for publication. / Graduation date: 2012