The proteasome is a large protein complex which is responsible for the majority of protein degradation in eukaryotes. Following FDA approval of the first proteasome inhibitor bortezomib for the treatment of multiple myeloma (MM) in 2003, there has been an increasing awareness of the significant therapeutic potential of proteasome inhibitors in the treatment of cancer. As of 2017, three proteasome inhibitors are approved for the treatment of MM but in clinical trials with patients bearing solid tumors these existing proteasome inhibitors have demonstrated poor results. Notably, all three FDA-approved proteasome inhibitors rely on the combination a peptide backbone and reactive electrophilic warhead to target the proteasome, and all three primarily target the catalytic subunits conferring the proteasome’s chymotrypsin-like (CT-L) activity. It is our hypothesis that compounds with non-peptidic structures, non-covalent and reversible modes of action, and unique selectivity profiles against the proteasome’s distinct catalytic subunits could have superior pharmacodynamic and pharmacokinetic properties and may bear improved activity against solid tumors relative to existing proteasome inhibitors. In an effort to discover such compounds we have employed an approach which combines computational drug screening methods with conventional screening and classic medicinal chemistry. Our efforts began with a computational screen performed in the lab of Dr. Chang-Guo Zhan. This virtual screen narrowed a library of over 300,000 drug-like compounds down to under 300 virtual hits which were then screened for proteasome inhibitory activity in an in vitro assay. Despite screening a relatively small pool of compounds, we were able to identify 18 active compounds. The majority of these hits were non-peptide in structure and lacked any hallmarks of covalent inhibition. The further development of one compound, a tri-substituted pyrazole, provided us with a proteasome inhibitor which demonstrated cytotoxic activity in a variety of human solid cancer cell lines as well as significant anti-tumor activity in a prostate cancer mouse xenograft model. We have also evaluated the in vitro pharmacokinetic properties of our lead compound and investigated its ability to evade cross-resistance phenomena in proteasome inhibitor-resistant cell lines. We believe that our lead compound as well as our drug discovery approach itself will be of interest and use to other researchers. We hope that this research effort may aid in the further development of reversible non-peptide proteasome inhibitors and may eventually deliver new therapeutic options for patients with difficult-to-treat solid tumors.

Since the introduction of the American Academy of Pediatrics Back to Sleep Campaign infants have not met the recommendation to “incorporate supervised, awake “prone play” in their infant’s daily routine to support motor development and minimize the risk of plagiocephaly”. Interventions are needed to increase infants’ tolerance for prone position and prone playtime to reduce the risk of plagiocephaly and motor delays. Associative learning is the ability to understand causal relationship between events. Operant conditioning is a form of associative learning that occurs by associating a behavior with positive or negative consequences. Operant conditions has been utilized to encourage behaviors such as kicking, reaching and sucking in infants by associating these behaviors with positive reinforcement. This dissertation is a compilation of three papers that each represent a study used to investigate a potential play based interventions to encourage prone motor skills in infants. The first paper describes a series of experiment used to develop the Prone Play Activity Center (PPAC) and experimental protocols used in the other studies. The purpose of the second study was to determine the feasibility of a clinical trial comparing usual care (low tech) to a high-tech intervention based on the principles of operant conditioning to increase tolerance for prone and improve prone motor skills. Ten infants participated in the study where parents of infants in the high tech intervention group (n=5) used the PPAC for 3 weeks to practice prone play. Findings from this study suggested the proposed intervention is feasible with some modifications for a future large-scale clinical trial. The purpose of the third study evaluated the ability of 3-6 months old infants to demonstrate AL in prone and remember the association learned a day later. Findings from this study suggested that a majority of infants demonstrated AL in prone with poor retention of the association, 24 hours later. Taken together these 3 papers provide preliminary evidence that a clinical trial of an intervention is feasible and that associative learning could be used to reinforce specific prone motor behaviors in the majority of infants.

The identification of disease genes (genes that when mutated cause human diseases) is an important and challenging problem. Proper diagnosis, prevention, as well as care for patients require an understanding of disease pathophysiology, which is best understood when the underlying causative gene(s) or genetic element(s) are identified. While the availability of the sequenced human genome helped to lead to the discovery of more than 1,900 disease genes, the rate of disease gene discovery is still occurring at a slow pace. The use of genetic linkage methods have successfully led to the identification of numerous disease genes. However, linkage studies are ultimately restricted by available meioses (clinical samples) which result in numerous candidate disease genes. This thesis addresses candidate gene prioritizations in disease gene discovery as applied toward a genetically heterogeneous disease known as Bardet-Biedl Syndrome (BBS). Specifically, the integration of various functional information and the development of a novel comparative genomic approach (Computational Orthologous Prioritization - COP) that led to the identification of BBS3 and BBS11. Functional data integration and application of the COP method may be helpful toward the identification of other disease genes.

Proteins are dynamic molecules capable of performing complex biological functions necessary for life. The impact of protein dynamics in the development of medicines is often understated. Science is only now beginning to unravel the numerous consequences of protein flexibility on structure and function. This thesis will encompass two case studies in developing small molecule inhibitors targeting flexible enzymes, and provide a thorough evaluation of their inhibitory mechanisms of action. The first case study focuses on caspases, a family of cysteine proteases responsible for executing the final steps of apoptosis. Consequently, they have been the subject of intense research due to the critical role they play in the pathogenesis of various cardiovascular and neurodegenerative diseases. A fragment-based screening campaign against human caspase-7 resulted in the identification of a novel series of allosteric inhibitors, which were characterized by numerous biophysical methods, including an X-ray co-crystal structure of an inhibitory fragment with caspase-7. The fragments described herein appear to have a significant impact on the substrate binding loop dynamics and the orientation of the catalytic Cys-His dyad, which appears to be the origin of their inhibition. This screening effort serves the dual purpose of laying the foundation for future medicinal chemistry efforts targeting caspase proteins, and for probing the allosteric regulation of this interesting class of hydrolases. The second case study focuses on glutamate racemase, another dynamic enzyme responsible for the stereoinversion of glutamate, providing the essential function of D-glutamate production for the crosslinking of peptidoglycan in all bacteria. Herein, I present a series of covalent inhibitors of an antimicrobial drug target, glutamate racemase. The application of covalent inhibitors has experienced a renaissance within drug discovery programs in the last decade. To leverage the superior potency and drug target residence time of covalent inhibitors, there have been extensive efforts to develop highly specific covalent modifications to reduce off-target liabilities. A combination of enzyme kinetics, mass spectrometry, and surface-plasmon resonance experiments details a highly specific 1,4-conjugate addition of a small molecule inhibitor with the catalytic Cys74 of glutamate racemase. Molecular dynamics simulations and quantum mechanics-molecular mechanics geometry optimizations reveal, with unprecedented detail, the chemistry of the conjugate addition. Two compounds from this series of inhibitors display antimicrobial potency comparable to β-lactam antibiotics, with significant activity against methicillin-resistant S. aureus strains. This study elucidates a detailed chemical rationale for covalent inhibition and provides a platform for the development of antimicrobials with a novel mechanism of action.

Many traditional approaches cease to be useful when the number of variables is large in comparison with the sample size. Penalized regression methods have proved to be an attractive approach, both theoretically and empirically, for dealing with these problems. This thesis focuses on the development of penalized regression methods for high-dimensional variable selection. The first part of this thesis deals with problems in which the covariates possess a grouping structure that can be incorporated into the analysis to select important groups as well as important members of those groups. I introduce a framework for grouped penalization that encompasses the previously proposed group lasso and group bridge methods, sheds light on the behavior of grouped penalties, and motivates the proposal of a new method, group MCP. The second part of this thesis develops fast algorithms for fitting models with complicated penalty functions such as grouped penalization methods. These algorithms combine the idea of local approximation of penalty functions with recent research into coordinate descent algorithms to produce highly efficient numerical methods for fitting models with complicated penalties. Importantly, I show these algorithms to be both stable and linear in the dimension of the feature space, allowing them to be efficiently scaled up to very large problems. In the third part of this thesis, I extend the idea of false discovery rates to penalized regression. The Karush-Kuhn-Tucker conditions describing penalized regression estimates provide testable hypotheses involving partial residuals. I use these hypotheses to connect the previously disparate elds of multiple comparisons and penalized regression, develop estimators for the false discovery rates of methods such as the lasso and elastic net, and establish theoretical results. Finally, the methods from all three sections are studied in a number of simulations and applied to real data from gene expression and genetic association studies.

Flexible enzymes are notoriously a bane to structure-based drug design and discovery efforts. This is because no single structure can accurately capture the vast array of conformations that exist in solution and many are subject to ligand-associated structural changes that are difficult to predict. Glutamate racemase (GR) – an antibiotic drug discovery target involved in cell wall biosynthesis – is one such enzyme that has eluded basic structure-based drug design and discovery efforts due to these flexibility issues. In this study, our focus is on overcoming the impediment of unpredictable ligand-associated structural changes in GR drug discovery campaigns. The flexibility of the GR active site is such that it is capable of accommodating ligands with very different structures. Though these ligands may bind to the same pocket, they may associate with quite dissimilar conformations where some are more favorable for complexation than others. Knowledge of these changes is invaluable in guiding drug discovery efforts, indicating which compounds selectively associate with more favorable conformations and are therefore better suited for optimization and providing starting structures to guide structure-based drug design optimization efforts. In this study, we develop a mutant GR possessing a genetically encoded non-natural fluorescent amino acid in a region remote from the active site whose movement has been previously observed to correlate with active site changes. With this mutant GR, we observe a differential fluorescence pattern upon binding of two structurally distinct competitive inhibitors known to associate with unique GR conformations – one to a favorable conformation with a smaller, less solvated active site and the other to an unfavorable conformation with a larger, more solvated active site. A concomitant computational study ascribes the source of this differential fluorescence pattern to ligand-associated conformational changes resulting in changes to the local environment of the fluorescent residue. Therefore, this mutant permits the elucidation of valuable structural information with relative ease by simply monitoring the fluorescence pattern resulting from ligand binding, which indicates whether the ligand has bound to a favorable or unfavorable conformation and offers insight into the general structure of this conformation.

Data containing large number of variables is becoming increasingly more common and sparsity inducing penalized regression methods, such the lasso, have become a popular analysis tool for these datasets due to their ability to naturally perform variable selection. However, quantifying the importance of the variables selected by these models is a difficult task. These difficulties are compounded by the tendency for the most predictive models, for example those which were chosen using procedures like cross-validation, to include substantial amounts of noise variables with no real relationship with the outcome. To address the task of performing inference on penalized regression models, this thesis proposes false discovery rate approaches for a broad class of penalized regression models. This work includes the development of an upper bound for the number of noise variables in a model, as well as local false discovery rate approaches that quantify the likelihood of each individual selection being a false discovery. These methods are applicable to a wide range of penalties, such as the lasso, elastic net, SCAD, and MCP; a wide range of models, including linear regression, generalized linear models, and Cox proportional hazards models; and are also extended to the group regression setting under the group lasso penalty. In addition to studying these methods using numerous simulation studies, the practical utility of these methods is demonstrated using real data from several high-dimensional genome wide association studies.

The primary research aim of this dissertation is to identify the strategies that best meet the information retrieval needs as expressed in the "e-discovery" scenario. This task calls for a high-recall system that, in response to a request for all available relevant documents to a legal complaint, effectively prioritizes documents from an enterprise document collection in order of likelihood of relevance. High recall information retrieval strategies, such as those employed for e-discovery and patent or medical literature searches, reflect high costs when relevant documents are missed, but they also carry high document review costs. Our approaches parallel the evaluation opportunities afforded by the TREC Legal Track. Within the ad hoc framework, we propose an approach that includes query field selection, techniques for mitigating OCR error, term weighting strategies, query language reduction, pseudo-relevance feedback using document metadata and terms extracted from documents, merging result sets, and biasing results to favor documents responsive to lawyer-negotiated queries. We conduct several experiments to identify effective parameters for each of these strategies. Within the relevance feedback framework, we use an active learning approach informed by signals from collected prior relevance judgments and ranking data. We train a classifier to prioritize the unjudged documents retrieved using different ad hoc information retrieval techniques applied to the same topic. We demonstrate significant improvements over heuristic rank aggregation strategies when choosing from a relatively small pool of documents. With a larger pool of documents, we validate the effectiveness of the merging strategy as a means to increase recall, but that sparseness of judgment data prevents effective ranking by the classifier-based ranker. We conclude our research by optimizing the classifier-based ranker and applying it to other high recall datasets. Our concluding experiments consider the potential benefits to be derived by modifying the merged runs using methods derived from social choice models. We find that this technique, Local Kemenization, is hampered by the large number of documents and the minimal number of contributing result sets to the ranked list. This two-stage approach to high-recall information retrieval tasks continues to offer a rich set of research questions for future research.

Ultra-deep RNA sequencing has become a powerful approach for genome-wide analysis of pre-mRNA alternative splicing. We develop MATS (Multivariate Analysis of Transcript Splicing), a Bayesian statistical framework for flexible hypothesis testing of differential alternative splicing patterns on RNA-Seq data. MATS uses a multivariate uniform prior to model the between-sample correlation in exon splicing patterns, and a Markov chain Monte Carlo (MCMC) method coupled with a simulation-based adaptive sampling procedure to calculate the P value and false discovery rate (FDR) of differential alternative splicing. Importantly, the MATS approach is applicable to almost any type of null hypotheses of interest, providing the flexibility to identify differential alternative splicing events that match a given user-defined pattern. We evaluated the performance of MATS using simulated and real RNA-Seq data sets. In the RNA-Seq analysis of alternative splicing events regulated by the epithelial-specific splicing factor ESRP1, we obtained a high RT-PCR validation rate of 86% for differential alternative splicing events with a MATS FDR of < 10%. Additionally, over the full list of RT-PCR tested exons, the MATS FDR estimates matched well with the experimental validation rate. Our results demonstrate that MATS is an effective and flexible approach for detecting differential alternative splicing from RNA-Seq data.

The initial aim of this project was to uncover novel cellulolytic organisms or enzymes from the diverse microbial source, activated sludge. Two isolation methods were used; either directly inoculating the sludge material onto filter paper as a carbon source, or using the Evolver� technology as an enrichment device. In both cases, as expected, cellulase activity was evident, however attributing this activity to one species was difficult in either case. This highlighted the complex interrelationships that existed between the many microorganisms present as the cellulosic carbon sources were degraded. In one instance, a Cellvibrio sp. was isolated. This genus of bacteria is known to possess both types of cellulase activity (exo- and endo- acting) and was therefore likely to contribute to the degradation of the cellulose. However, the isolate, once purified, did not display significant cellulolytic ability as compared to the unpurified consortium of microorganisms. Therefore, in each case, microorganisms responsible for the cellulolytic activity were not uncovered. It was suspected that the microorganisms responsible for some of the cellulolytic activity were protists. During the isolation of microorganisms, an Aeromonas sp. bearing the novel phenotype (for this genus) of CMCase activity was isolated. This activity was at first suspected to contribute to the degradation of the filter paper that was seen during isolation. However, tests with the pure isolate suggested that the Aeromonas sp. CMCase was not used for cellulose catabolism. Ironically, the enzyme may instead function in the production of a cellulose-like exopolysaccharide by the bacterium. Part of a cellulose synthase operon was found in the genome of the Aeromonas sp. isolate, including a gene coding for an endoglucanase that gives a predicted molecular weight enzyme similar to the 39 kDa CMCase purified from the bacterium. The CMCase enzyme, operating as part of of a synthetic operon is expected to be important in terms of the biofilm forming ability of this Aeromonas strain. Such capabilities of the bacterium were investigated here, including observing motility behaviour of the organism on agar surfaces. Studying the biofilm forming ability of this genus in general will be important in understanding how the fish and human pathogens persist in aquatic environments