Septinas compreendem uma conservada família de proteínas de ligação a nucleotídeo de guanina e formação de heterofilamentos. Em termos estruturais, elas possuem uma organização comum: um domínio GTPase central, uma região N-terminal e um domínio C-terminal, este último é predito para formar estruturas em coiled coil. Atualmente, o heterocomplexo de septinas humanas (SEPT2/SEPT6/SEPT7) mais bem caracterizado revela a importância do domínio GTPase na formação do filamento, todavia a ausência de densidade eletrônica para os domínios C-terminais faz com que sua função permaneça obscura. Estudos com septinas de mamíferos, e de outros organismos como C. elegans e S. cerevisea sugerem que alguns grupos de septinas (por exemplo, II e IV em mamíferos) interagem através de seus domínios C-terminais, e estes poderiam atuar de modo determinante para a montagem correta do filamento. Assim, o presente projeto objetivou estudar a afinidade homo/heterotípicas para os domínios C-terminais das septinas humanas dos grupos II (SEPT6C/8C/10C/11C) e IV (SEPT7C), investigando se esses domínios contribuem para preferência das septinas interagirem com proteínas de grupos distintos durante a formação do heterofilamento. Os domínios C-terminais foram expressos em E. coli e purificados. Foram conduzidos estudos de ultracentrifugação analítica e espectropolarimetria de dicroísmo circular, que permitiram identificar maior afinidade e estabilidade da associação heterotípica comparada à homotípica. Foram obtidas constantes de dissociação aparente para homodímeros em torno de baixo µM, enquanto que para heterodímeros os dados já existentes no grupo revelaram constante de dissociação na ordem de nM. Para entender os fatores no nível atômico responsáveis pela significativa predileção na interação entre os domínios C-terminais dos grupos II e IV foram realizados estudos utilizando modelagem e análise das sequências primárias. As análises sugerem a presença de um alto número de resíduos carregados na posição a do coiled coil como responsável pela seletividade. Consequentemente, o heterodímero seria favorecido em virtude do menor efeito repulsivo proveniente do intercalamento dos resíduos carregados em a. Desse modo, os resultados indicaram a atuação decisiva ou cooperativa dos domínios C-terminais na organização preferencial das septinas durante a formação do filamento, favorecendo a interface NC entre septinas dos grupos II e IV. / Septins comprise a conserved protein family that binds guanidine nucleotide and forms heterofilaments. In structural terms they have a common organization: a central GTPase domain, a N-terminal domain and a C-terminal domain, this last one is predicted to form coiled coil structures. Currently, the human septin heterocomplex best characterized (SEPT2/SEPT6/SEPT7) reveals the importance of the GTPase domain in filament assembly, however the absence of electron density for the C-terminal domains makes its function still unknown. Studies with mammals septins, and of others organisms like C. elegans and S. cerevisea suggests that some septins groups (e.g. II e IV in mammals) interact via its C-terminal domains and this could act in a determinative way to correct filament assembly. In this way, this project aimed to study the homo/heterotypical affinity for the C-terminal domains of human septins belonging to groups II (SEPT6C/8C/10C/11C) e IV (SEPT7C), investigating whether this domain contributes with the preference of septins to interact with proteins of different groups during assembly of the heterofilament. The C-terminal domains were expressed in E. coli and purificated. It was carried out studies using analytical ultracentrifugation and circular dichroism spectropolarimetry tecniques which allowed identification of major affinity and stability in the heterotypical association compared to homotypical. It was measured apparent dissociation constants for homodimers of low µM range while for heterodimers our group\'s data shows dissociation constants in the nM range. To understand at atomic level the factors responsible for this significant preference in the C-terminal domains interaction between groups II and IV was performed molecular modelling studies and analysis of the primary sequence. These analysis suggests the presence of a high number of charged residues in position a of the coiled coil as responsible for selectivity. Consequently, the heterodimer would be therefore favoured because of the minor repulsive effect coming from the staggered of charged residues in a. Thus, these results indicate the crucial or cooperative action of C-terminal domains in preferential organization of septins during filament assembly, favouring the NC interface between septins of groups II and IV.

With new tests being developed and marketed, the comparison of the diagnostic accuracy of two continuous-scale diagnostic tests are of great importance. Comparing the partial areas under the receiver operating characteristic curves (pAUC) is an effective method to evaluate the accuracy of two diagnostic tests. In this thesis, we study the semi-parametric inference for the difference between two pAUCs. A normal approximation for the distribution of the difference between two pAUCs has been derived. The empirical likelihood ratio for the difference between two pAUCs is defined and its asymptotic distribution is shown to be a scaled chi-quare distribution. Bootstrap and empirical likelihood based inferential methods for the difference are proposed. We construct five confidence intervals for the difference between two pAUCs. Simulation studies are conducted to compare the finite sample performance of these intervals. We also use a real example as an application of our recommended intervals.

Compare the accuracy of two continuous-scale tests is increasing important when a new test is developed. The traditional approach that compares the entire areas under two Receiver Operating Characteristic (ROC) curves is not sensitive when two ROC curves cross each other. A better approach to compare the accuracy of two diagnostic tests is to compare the areas under two ROC curves (AUCs) in the interested specificity interval. In this thesis, we have proposed bootstrap and empirical likelihood (EL) approach for inference of the difference between two partial AUCs. The empirical likelihood ratio for the difference between two partial AUCs is defined and its limiting distribution is shown to be a scaled chi-square distribution. The EL based confidence intervals for the difference between two partial AUCs are obtained. Additionally we have conducted simulation studies to compare four proposed EL and bootstrap based intervals.

Septinas compreendem uma conservada família de proteínas de ligação a nucleotídeo de guanina e formação de heterofilamentos. Em termos estruturais, elas possuem uma organização comum: um domínio GTPase central, uma região N-terminal e um domínio C-terminal, este último é predito para formar estruturas em coiled coil. Atualmente, o heterocomplexo de septinas humanas (SEPT2/SEPT6/SEPT7) mais bem caracterizado revela a importância do domínio GTPase na formação do filamento, todavia a ausência de densidade eletrônica para os domínios C-terminais faz com que sua função permaneça obscura. Estudos com septinas de mamíferos, e de outros organismos como C. elegans e S. cerevisea sugerem que alguns grupos de septinas (por exemplo, II e IV em mamíferos) interagem através de seus domínios C-terminais, e estes poderiam atuar de modo determinante para a montagem correta do filamento. Assim, o presente projeto objetivou estudar a afinidade homo/heterotípicas para os domínios C-terminais das septinas humanas dos grupos II (SEPT6C/8C/10C/11C) e IV (SEPT7C), investigando se esses domínios contribuem para preferência das septinas interagirem com proteínas de grupos distintos durante a formação do heterofilamento. Os domínios C-terminais foram expressos em E. coli e purificados. Foram conduzidos estudos de ultracentrifugação analítica e espectropolarimetria de dicroísmo circular, que permitiram identificar maior afinidade e estabilidade da associação heterotípica comparada à homotípica. Foram obtidas constantes de dissociação aparente para homodímeros em torno de baixo µM, enquanto que para heterodímeros os dados já existentes no grupo revelaram constante de dissociação na ordem de nM. Para entender os fatores no nível atômico responsáveis pela significativa predileção na interação entre os domínios C-terminais dos grupos II e IV foram realizados estudos utilizando modelagem e análise das sequências primárias. As análises sugerem a presença de um alto número de resíduos carregados na posição a do coiled coil como responsável pela seletividade. Consequentemente, o heterodímero seria favorecido em virtude do menor efeito repulsivo proveniente do intercalamento dos resíduos carregados em a. Desse modo, os resultados indicaram a atuação decisiva ou cooperativa dos domínios C-terminais na organização preferencial das septinas durante a formação do filamento, favorecendo a interface NC entre septinas dos grupos II e IV. / Septins comprise a conserved protein family that binds guanidine nucleotide and forms heterofilaments. In structural terms they have a common organization: a central GTPase domain, a N-terminal domain and a C-terminal domain, this last one is predicted to form coiled coil structures. Currently, the human septin heterocomplex best characterized (SEPT2/SEPT6/SEPT7) reveals the importance of the GTPase domain in filament assembly, however the absence of electron density for the C-terminal domains makes its function still unknown. Studies with mammals septins, and of others organisms like C. elegans and S. cerevisea suggests that some septins groups (e.g. II e IV in mammals) interact via its C-terminal domains and this could act in a determinative way to correct filament assembly. In this way, this project aimed to study the homo/heterotypical affinity for the C-terminal domains of human septins belonging to groups II (SEPT6C/8C/10C/11C) e IV (SEPT7C), investigating whether this domain contributes with the preference of septins to interact with proteins of different groups during assembly of the heterofilament. The C-terminal domains were expressed in E. coli and purificated. It was carried out studies using analytical ultracentrifugation and circular dichroism spectropolarimetry tecniques which allowed identification of major affinity and stability in the heterotypical association compared to homotypical. It was measured apparent dissociation constants for homodimers of low µM range while for heterodimers our group\'s data shows dissociation constants in the nM range. To understand at atomic level the factors responsible for this significant preference in the C-terminal domains interaction between groups II and IV was performed molecular modelling studies and analysis of the primary sequence. These analysis suggests the presence of a high number of charged residues in position a of the coiled coil as responsible for selectivity. Consequently, the heterodimer would be therefore favoured because of the minor repulsive effect coming from the staggered of charged residues in a. Thus, these results indicate the crucial or cooperative action of C-terminal domains in preferential organization of septins during filament assembly, favouring the NC interface between septins of groups II and IV.

Doctor of Philosophy / Department of Statistics / Wei-Wen Hsu / The classification methods based on a linear combination of multiple biomarkers have been widely used to improve the accuracy in disease screening and diagnosis. However, it is seldom to include covariates such as gender and age at diagnosis into these classification procedures. It is known that biomarkers or patient outcomes are often associated with some covariates in practice, therefore the inclusion of covariates may further improve the power of prediction as well as the classification accuracy. In this study, we focus on the classification methods for multiple biomarkers adjusting for covariates. First, we proposed a covariate-adjusted classification model for multiple cross-sectional biomarkers. Technically, it is a two-stage method with a parametric or non-parametric approach to combine biomarkers first, and then incorporating covariates with the use of the maximum rank correlation estimators. Specifically, these parameter coefficients associated with covariates can be estimated by maximizing the area under the receiver operating characteristic (ROC) curve. The asymptotic properties of these estimators in the model are also discussed. An intensive simulation study is conducted to evaluate the performance of this proposed method in finite sample sizes. The data of colorectal cancer and pancreatic cancer are used to illustrate the proposed methodology for multiple cross-sectional biomarkers. We further extend our classification method to longitudinal biomarkers. With the use of a natural cubic spline basis, each subject's longitudinal biomarker profile can be characterized by spline coefficients with a significant reduction in the dimension of data. Specifically, the maximum reduction can be achieved by controlling the number of knots or degrees of freedom in the spline approach, and its coefficients can be obtained by the ordinary least squares method. We consider each spline coefficient as ``biomarker'' in our previous method, then the optimal linear combination of those spline coefficients can be acquired using Stepwise method without any distributional assumption. Afterward, covariates are included by maximizing the corresponding AUC as the second stage. The proposed method is applied to the longitudinal data of Alzheimer's disease and the primary biliary cirrhosis data for illustration. We conduct a simulation study to assess the finite-sample performance of the proposed method for longitudinal biomarkers.

Data mining involves the use of data analysis tools to discover previously unknown, valid patterns and relationships from large amounts of data stored in databases, data warehouses, or other information repositories. Feature selection is an important preprocessing step of data mining that helps increase the predictive performance of a model. The main aim of feature selection is to choose a subset of features with high predictive information and eliminate irrelevant features with little or no predictive information. Using a single feature selection technique may generate local optima. In this thesis we propose an ensemble approach for feature selection, where multiple feature selection techniques are combined to yield more robust and stable results. Ensemble of multiple feature ranking techniques is performed in two steps. The first step involves creating a set of different feature selectors, each providing its sorted order of features, while the second step aggregates the results of all feature ranking techniques. The ensemble method used in our study is frequency count which is accompanied by mean to resolve any frequency count collision. Experiments conducted in this work are performed on the datasets collected from Kent Ridge bio-medical data repository. Lung Cancer dataset and Lymphoma dataset are selected from the repository to perform experiments. Lung Cancer dataset consists of 57 attributes and 32 instances and Lymphoma dataset consists of 4027 attributes and 96 ix instances. Experiments are performed on the reduced datasets obtained from feature ranking. These datasets are used to build the classification models. Model performance is evaluated in terms of AUC (Area under Receiver Operating Characteristic Curve) performance metric. ANOVA tests are also performed on the AUC performance metric. Experimental results suggest that ensemble of multiple feature selection techniques is more effective than an individual feature selection technique.

In diagnostic test studies, one crucial task is to evaluate the diagnostic accuracy of a test. Currently, most studies focus on the Receiver Operating Characteristics Curve and the Area Under the Curve. On the other hand, the Youden index, widely applied in practice, is another comprehensive measurement for the performance of a diagnostic test. For a continuous-scale test classifying diseased and non-diseased groups, finding the Youden index of the test is equivalent to maximize the sum of sensitivity and specificity for all the possible values of the cut-point. This dissertation concentrates on statistical inferences for the Youden index. First, an auxiliary tool for the Youden index, called the diagnostic curve, is defined and used to evaluate the diagnostic test. Second, in the paired-design study to assess the diagnostic accuracy of two biomarkers, the difference in paired Youden indices frequently acts as an evaluation standard. We propose an exact confidence interval for the difference in paired Youden indices based on generalized pivotal quantities. A maximum likelihood estimate-based interval and a bootstrap-based interval are also included in the study. Third, for certain diseases, an intermediate level exists between diseased and non-diseased status. With such concern, we define the Youden index for three ordinal groups, propose the empirical estimate of the Youden index, study the asymptotic properties of the empirical Youden index estimate, and construct parametric and nonparametric confidence intervals for the Youden index. Finally, since covariates often affect the accuracy of a diagnostic test, therefore, we propose estimates for the Youden index with a covariate adjustment under heteroscedastic regression models for the test results. Asymptotic properties of the covariate-adjusted Youden index estimators are investigated under normal error and non-normal error assumptions.

Customer targeting is the key to the success of bank telemarketing. To compare the flexible discriminant analysis and the logistic regression in customer targeting, a survey dataset from a Portuguese bank was used. For the flexible discriminant analysis model, the backward elimination of explanatory variables was used with several rounds of manual re-defining of dummy variables. For the logistic regression model, the automatic stepwise selection was performed to decide which explanatory variables should be left in the final model. Ten-fold stratified cross validation was performed to estimate the model parameters and accuracies. Although employing different sets of explanatory variables, the flexible discriminant analysis model and the logistic regression model show equally satisfactory performances in customer classification based on the areas under the receiver operating characteristic curves. Focusing on the predicted “right” customers, the logistic regression model shows slightly better classification and higher overall correct prediction rate.

Species distribution models (SDM) are widely used tool in biogeography, macroecology and nature conservation. With gradual development, it has become an important means used by, for example, in determining the potentially threatened locations by invasive species, or studying the impact of climate change on biodiversity. With the progressive development it becomes obvious that one of the major factors limiting the species distribution modelling are input data. The presence data are most readily available, but they suffer from an uneven collection, for example, with a predominance of records in easily accessible locations. The aim of this work is to show, that popular climate filtering of presence data input, in order to eliminate uneven sampling, affects the final model in a negative way. For this purpose, there were virtual sorts of different species and different prevalence of recorded occurrences on the territory of the Iberian Peninsula generated. Subsequently, species distribution models with and without climate filters were created by using Maxent. They were evaluated by AUC. The difference between virtual reality, which is presented to the suitability of the virtual species, and the resulting model was tested by paired T test. Comparison of the AUC confirmed that the species distribution models based on climate filtering have better discriminative ability. However, it only points to the skilful work with the selected sample bias that already does not reflect reality. In contrast, comparison of the differences between virtual reality and the models with and without climate filtering using a paired T test shows greater congruence between unfiltered models and virtual reality. Thus it was proved that the climate filtering does not lead to higher validity species distribution models.

Indiana University-Purdue University Indianapolis (IUPUI) / Forensic DNA phenotyping is an up and coming area in forensic DNA analyses that enables the prediction of physical appearance of an individual from DNA left at a crime scene. At present, there has been substantial work performed in understanding what genes/markers are required to produce a reliable prediction of categorical eye and hair color from the DNA of an individual of interest. These pigmentation markers (variants from HERC2, OCA2, TYR, SLC24A4, SLC45A2, IRF4 to name a few) are at the core of several prediction systems for eye and hair color such as IrisPlex, HIrisPlex, and the Snipper 2.5 suite. The contribution of these markers towards prediction in most cases however, only factors in an independent effect and do not take into account potential interactions or epistasis in the production of the final phenotypic color. Epistasis is a phenomenon that occurs when a gene’s effect relies on the presence of ‘modifier genes’, and can display different effects (enhance/repress a particular color) in genotype combinations rather than individually. In an effort to detect such epistatic interactions and their influence on hair color prediction models, for this current study, 872 individuals were genotyped at 61 associative and predictive pigmentation markers from several diverse population subsets. Individuals were phenotypically evaluated for eye and hair color by three separate independent assessments. Several analyses were performed using statistical approaches such as multifactor dimensionality reduction (MDR) for example, in an effort to detect if there are any SNP- SNP epistatic interactions present that could potentially enhance eye and hair color prediction model performances. The ultimate goal of this study was to assess what SNP-SNP combinations amongst these known pigmentation genes should be included as an additional variable in future prediction models and how much they can potentially enhance overall pigmentation prediction model performance. The second part of the project involved the analyses of several differentially expressed candidate genes between different hair color follicles of the same individual using quantitative Real Time PCR. We looked at 26 different genes identified through a concurrent non-human primate study being performed in the laboratory. The purpose of this study was to gain more insight on the level of differentially expressed mRNA between different hair color follicles within the same human individual. Data generated from this part of the project will act as a pilot study or ‘proof of principle’ on the mRNA expression of several pigmentation associated genes on individual beard hair of varying phenotypic colors. This analysis gives a first glimpse at expression levels that remain constant or differentiate between hairs of the same individual, therefore limiting the contribution of individual variation.