This thesis is about multiple hypothesis testing and its relation to the P-value. In Chapter 1, the methodologies of hypothesis testing among the three inference schools are reviewed. Jeffreys, Fisher, and Neyman advocated three different approaches for testing by using the posterior probabilities, P-value, and Type I error and Type II error probabilities respectively. In Berger's words ``Each was quite critical of the other approaches." Berger proposed a potential methodological unified conditional frequentist approach for testing. His idea is to follow Fisher in using the P-value to define the strength of evidence in data and to follow Fisher's method of conditioning on strength of evidence; then follow Neyman by computing Type I and Type II error probabilities conditioning on strength of evidence in the data, which equal the objective posterior probabilities of the hypothesis advocated by Jeffreys. Bickis proposed another estimate on calibrating the null and alternative components of the distribution by modeling the set of P-values as a sample from a mixed population composed of a uniform distribution for the null cases and an unknown distribution for the alternatives. For tackling multiplicity, exploiting the empirical distribution of P-values is applied. A variety of density estimators for calibrating posterior probabilities of the null hypothesis given P-values are implemented. Finally, a noninterpolatory and shape-preserving estimator based on B-splines as smoothing functions is proposed and implemented.

Systems biologists use interaction graphs to model the behaviour of biological systems at the molecular level. In an iterative process, such biologists observe the reactions of living cells under various experimental conditions, view the results in the context of the interaction graph, and then propose changes to the graph model. These graphs represent dynamic knowledge of the biological system being studied and evolve as new insight is gained from the experimental data. While numerous graph layout and drawing packages are available, these tools did not fully meet the needs of our immunologist collaborators. In this thesis, we describe the data display needs of these immunologists and translate these needs into visual encoding decisions. These decisions led us to create Cerebral, a system that uses a biologically guided graph layout and incorporates experimental data directly into the graph display. Our graph layout algorithm uses simulated annealing with constraints, optimized with a uniform grid to have an expected runtime of o(E/V). Small multiple views of different experimental conditions and a measurement-driven parallel coordinates view enable correlations between experimental conditions to be analyzed at the same time that the measurements are viewed in the graph context. This combination of coordinated views allows the biologist to view the data from many different perspectives simultaneously. To illustrate the typical analysis tasks performed, we analyze two datasets using Cerebral. Based on feedback from our collaborators, we conclude that Cerebral is a valuable tool for analyzing experimental data in the context of an interaction graph model.

This thesis is about multiple hypothesis testing and its relation to the P-value. In Chapter 1, the methodologies of hypothesis testing among the three inference schools are reviewed. Jeffreys, Fisher, and Neyman advocated three different approaches for testing by using the posterior probabilities, P-value, and Type I error and Type II error probabilities respectively. In Berger's words ``Each was quite critical of the other approaches." Berger proposed a potential methodological unified conditional frequentist approach for testing. His idea is to follow Fisher in using the P-value to define the strength of evidence in data and to follow Fisher's method of conditioning on strength of evidence; then follow Neyman by computing Type I and Type II error probabilities conditioning on strength of evidence in the data, which equal the objective posterior probabilities of the hypothesis advocated by Jeffreys. Bickis proposed another estimate on calibrating the null and alternative components of the distribution by modeling the set of P-values as a sample from a mixed population composed of a uniform distribution for the null cases and an unknown distribution for the alternatives. For tackling multiplicity, exploiting the empirical distribution of P-values is applied. A variety of density estimators for calibrating posterior probabilities of the null hypothesis given P-values are implemented. Finally, a noninterpolatory and shape-preserving estimator based on B-splines as smoothing functions is proposed and implemented.

Multiple comparisons and selection procedures are commonly studied in research and employed in application. Clinical trial is one of popular fields to which the subject of multiple comparisons is extensively applied. Based on the Federal Food, Drug, and Cosmetic Act, drug manufacturers need to not only demonstrate safety of their drug products but also establish effectiveness by substantial evidence in order to obtain marketing approval. However, the problem of error inflation occurs when there are more than two groups to compare with at the same time. How to design a test procedure with high power while controlling type I error becomes an important issue. The treatment with the largest population mean is considered to be the best one in the study. Potentially the best treatments can receive increased resources and further investigation by excluding clearly inferior treatments. Hence, a small number of possibly the best treatments is preferred. This thesis focuses on the problem of eliminating the less effective treatments among three in clinical trials. The goal is to increase the ability to identify any inferior treatment providing that the probability of excluding any best treatment is guaranteed to be less than or equal to alpha. A step-down procedure is applied to solve the problem. The general step-down procedure with fixed thresholds is conservative in our problem. The test is not efficient in rejecting the less effective treatments. We propose two methods with sharper thresholds to improve current procedures and construct a subset containing strictly inferior treatments. The first method, the restricted parameter space approach, is designed for the scenario when prior information about range of treatment means is known. The second method, the step-down procedure with feedback, utilizes observations to modify the threshold and controls error rate for the whole parameter space. The new procedures have greater ability to detect more inferior treatments than the standard procedure. In addition, type I error is also controlled under mild violation of the assumptions demonstrated by simulation.

Multiple sclerosis (MS) is the most common primary demyelinating disease of the central nervous system in humans. Type I interferons are most frequently used to treat MS. However, their main mechanism of action remains elusive. Various biomarkers have been investigated for their ability to assess treatment efficacy, but results are often confounding due to differences in experimental design and variation in individual physiology. In fact, not all MS patients respond to IFN therapy and a significant number suffer severe negative side effects and must cease treatment. Thus, alternative therapeutics that offer less cytotoxicity and greater efficacy are a major objective of research. This dissertation evaluated a novel type I interferon, interferon-tau (IFNT), and its ability to attenuate Theiler’s virus-induced-demyelinating disease (TVID), a mouse model of MS. In this model, viral infection with the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV) is the initiating factor leading to demyelination of the CNS. It was hypothesized that IFNT would: 1) provide therapeutic benefit as witnessed by a stabilization of clinical score, a decrease in CNS inflammation, and a decrease in CNS demyelination, and 2) shift the immune profile from a Th1 to a Th2 response. Once mice developed chronic disability, IFNT treatment began. This novel IFN was delivered in an innovative way: encapsulation (eIFNT) in an alginate polymer, which allowed for slow and sustained release. eIFNT was delivered by a 100 μl intraperitoneal injection (i.p.) containing 1.4M U of IFNT once every two weeks for 8 weeks. Mice were clinically scored weekly and BeAn-eIFNT mice demonstrated a decrease in clinical score. Bright field microscopy was used to evaluate CNS tissues where a decrease in demyelination and inflammation was noted in BeAn-eIFNT-treated mice. Ex vivo stimulation of virus-specific lymphocytes revealed an increase in both T helper 1 (Th1) and T helper 2 (Th2) cytokine production. Specifically, TNFA was produced at very high levels by splenocytes from BeAneIFNT mice in response to UV-inactivated BeAn alone and in the presence of IFNT when compared to BeAn-eMOPS mice under the same conditions. IFNG was produced at elevated levels from the splenocytes of BeAn-eIFNT mice versus BeAn-eMOPS mice when stimulated in vitro with UV-inactivated Bean and with BeAn in the presence of IFNT. IL-2 was produced at moderately elevated levels from the splenocytes of BeAn-eIFNT mice versus BeAn-eMOPS mice when stimulated in vitro with UV-inactivated Bean. Il-2 was elevated to a statistically significant level (p<0.05) from BeAn-eIFNT mouse splenocytes when stimulated with BeAn in the presence of IFNT when compared to BeAn-eMOPS mice and IL-10 was produced at elevated levels by splenocytes from BeAn-eIFNT mice versus that produced from BeAn-eMOPS mouse splenocytes in response to UV-inactivated BeAn alone and in the presence of IFNT. Quantification of T regulatory (Treg) cells in the spleen of eIFNT vs. eMOPS mice and blood of eIFNT vs. eMOPS mice revealed no difference between the two groups. There was no statistical difference in virus-specific serum antibodies at the pretreatment time point noted in the OD readings of eIFNT mice at a dilution of 1/200 compared to the eMOPS mice. A modest decrease in the OD values at the 1/200 dilution were noted in the eIFNT mice compared to the eMOPS mice, but this difference was not significant. Antibody secreting cells (ASCs) from eIFNT mice versus eMOPS mice were slightly lower in the spleen and brains whereas there was a slight increase in ASCs from the spinal cord of eIFNT mice when compared to those from eMOPS mice. Altogether, the results support efficacy of the eIFNT treatment in the mice with TVID. Actual mechanisms of disease attenuation remain elusive at this time as mice exhibited an increase in certain Th1 and Th2 cytokines rather than the hypothesized shift from a Th1 to a Th2 immune profile. Likewise, mice exhibited a modest decrease in virus specific antibodies as well as the number virus-specific ASCs which also refute the hypothesized increase in these values. A remarkable finding was the fact that immune cells derived from eIFNT treated mice appeared to be divided into two distinct types of biological responders although all of the mice responded to the in vivo treatment with a decrease in disease severity. It is hypothesized that this difference is a reflection of individual genetic variability in response to immune modulation which is surprising owing to the fact that the animals used for these studies are in-bred and considered to be as identical genetically as is feasible in a population of animals. Obviously, immune modulation can proceed through different mechanisms and still provide the desired result of a decrease in disease severity. However, this reality creates an added level of difficulty when one is trying to interpret biological data in order to determine whether a therapeutic regimen is efficacious within a patient population.

Unmanned Vehicles (UVs) are developed for several civil and military applications. For these applications, there is a need for multiple vehicles with different capabilities to visit and monitor a set of given targets. In such scenarios, routing problems arise naturally where there is a need to plan paths in order to optimally use resources and time. The focus of this thesis is to address a basic optimization problem that arises in this setting. We consider a routing problem where some targets have to be visited by specific vehicles. We approach this problem by dividing the routing into two sub problems: partitioning the targets while satisfying vehicle target constraints and sequencing. We solve the partitioning problem with the help of a minimum spanning tree algorithm. We use 3 different approaches to solve the sequencing problem; namely, the 2 approximation algorithm, Christofide's algorithm and the Lin - Kernighan Heuristic (LKH). The approximation algorithms were implemented in MATLAB. We also developed an integer programming (IP) model and a relaxed linear programming (LP) model in C with the help of Concert Technology for CPLEX, to obtain lower bounds. We compare the performance of the developed approximation algorithms with both the IP and the LP model and found that the heuristic performed very well and provided the better quality solutions as compared to the approximation algorithms. It was also found that the approximation algorithms gave better solutions than the apriori guarantees.

Many tests have been developed for comparing means in a two-sample scenario. Microarray experiments lead to thousands of such comparisons in a single study. Several multiple testing procedures are available to control experiment-wise error or the false discovery rate. In this dissertation, individual two-sample tests are compared based on accuracy, correctness, and power. Four multiple testing procedures are compared via simulation, based on data from the lab of Dr. Rajesh Miranda. The effect of sample size on power is also carefully examined. The two sample t-test followed by the Benjamini and Hochberg (1995) false discovery rate controlling procedure result in the highest power.

Thèse doctorat : Médecine : Université de Nantes : 2005. / Bibliogr. 143-159 f. [266 réf.].

Thèse : Mathématiques : Université, Faculté des sciences de Paris : 1843. / Titre provenant de l'écran-titre.

Reproduction de : Thèse de doctorat : Médecine. Immunologie : Nantes : 2007. / Bibliogr.