22689 research outputs found
Sort by
Assessing genetic predictors of antibiotic resistance phenotypes in wastewater Pseudomonas isolates
No full textThe escalating threat of antimicrobial resistance (AMR), recognized by the World Health Organization as a major global health challenge, underscores the urgency of enhancing resistance detection methods. Advances in sequencing technology have not only improved the identification of AMR genes but also provided the possibility of streamlining the diagnosis and treatment of infections without the need for extensive culturing or isolation of microbes. These advancements also facilitate the global monitoring of AMR genes through the analysis of environmental isolates. However, the performance of gene identification tools for resistance phenotype prediction remains underexplored.
This study investigates the genotype-phenotype relationship in Pseudomonas strains collected from wastewater influent in the Baltimore region, a source of AMR genes relevant to human health. To determine whether drug resistance pattern is affected by evolutionary distance, we correlated the 16S rRNA gene sequence distance of 220 Pseudomonas with the similarity of resistance profiles against 13 antibiotics. Linear regression analysis revealed a significant, although very weak, negative relationship between phenotype similarity and evolutionary distance. This suggests that organisms with greater evolutionary distances exhibit slightly less antibiotic phenotype similarity, although the relationship is not robust enough to predict resistance profiles based solely on evolutionary distance. Analysis of very closely related isolates and their short evolutionary distances could slightly diminish this relationship.
Subsequently, we examined the concordance between in silico predictions of phenotypic resistance and the presence of resistance genes in the genome using metrics such as precision, specificity, accuracy, and the Matthews Correlation Coefficient. Our analysis revealed that higher gene similarity and longer sequence overlap with the database generally lead to better performance. Notably, gene identities above 94% and percentage lengths of reference sequence above 72% yields the highest levels of precision and specificity, which are statistically significant. These findings aid clinicians in making informed decisions about medication choices, including which antibiotics to use for susceptible strains and which to avoid for resistant strains. However, the translation from gene presence to gene expression and phenotype adds complexity to these predictions. The prevalence of efflux pumps, which confer resistance to multiple antibiotics, further complicates the determination of drug-specific phenotypes
Integrative and Transfer Learning Methods for Disparate Data with Applications in Single-cell Genomics and Statistical Genetics
No full textIn recent research of genomics and genetics, we face the challenges how to perform data analysis better by incorporating more information either internally from dataset itself or externally from other studies. My thesis will discuss about three approaches to addressing this problem.
The first part reviews existing benchmark works and introduces a new method, \texttt{mixhvg}, for selecting highly variable genes in single-cell RNA-sequencing. This process is vital due to the intrinsic characteristics of single-cell RNA-sequencing. Our work not only fills the gap in comprehensive benchmarks for selecting optimal methods but also proposes \texttt{mixhvg}, a hybrid approach that enhances performance robustly.
The second part builds on the first, examining the effects of highly variable gene selection on downstream analysis, specifically visualization. We find that selected genes depend on data, where local structures benefit from corresponding local gene selections. To illustrate local patterns more effectively, we propose an adaptive method, \texttt{SAVIS}, which integrates with \texttt{mixhvg} to further improve outcomes.
The third part discusses heterogeneous transfer learning for disparate datasets with unmatched feature sets. We address this by exploring transfer learning between two data types to construct high-dimensional generalized linear models. Here, the primary dataset has a smaller sample size but a comprehensive set of variables of interest, while the external dataset is larger but feature-limited. Our proposed HTL-GMM method utilizes the generalized method of moments (GMM) to enhance both prediction accuracy and post-selection inference effectively.
Overall, this thesis focuses on integrative and transfer learning methods applicable to single-cell genomics and genetic data, aiming to advance analytical capabilities in these fields
REDUCTION OF POSITION ERROR IN INDUSTRIAL CT SCANNING APPLICATION TO IMPROVE RECONSTRUCTION ACCURACY
No full textOver the years following the invention of CT scanners, its applications have broadened beyond its medical use cases and into vital industrial applications. Industrial CT scanning coupled with powerful software lends itself a vital tool for engineers to quickly diagnose part defects, non-destructively identify porosity, identify internal failures in parts and verify dimensions, amongst other applications.
One company that has emerged to provide this integral tool is called Lumafield. Unlike a medical CT scanner, Lumafield’s Neptune relies on a motion system to position parts in front of the X-ray source and rotate incrementally as x-ray radiographs are captured. In the design of these scanners, variables that affect reconstruction of parts must be properly controlled. One variable in particular that is integral to control is the Source-to-Object Distance (SOD). This parameter directly drives the scale factor used to convert radiographs of objects into 3D reconstructions and is dictated by the x-axis position of the motion system.
Testing of the SOD accuracy indicated that the current scanners are vulnerable to motion flatness errors. To address these errors, multiple experiments were conducted to determine that rail straightness profile errors are root cause to these flatness errors. The best solution to mitigate this problem is to mount a counter rail with a similar straight error as the main rail to the back of the x-axis assembly. This method boasts multiple benefits from a mechanical design perspective and manufacturing efficiency standpoint.
To integrate the solution into production, a rail profile measuring device was designed to identify the rail profile quickly and simply. Thus, the goal of this project and thesis is to design the measuring device and discuss the verification processes implemented to prove the efficacy of the counter rail solution
Navigating Beyond the Standard Model: A Miscellany
Get PDFThe Standard Model successfully codifies the fundamental particles and interactions that are amenable to observation. Together with the general theory of relativity and LambdaCDM cosmology, it provides a compelling account of the structure and evolution of physical systems throughout (and including) the Universe. However, this picture is necessarily incomplete, leaving unanswered a number of theoretical puzzles alongside such basic questions as the nature of dark matter. The space of possibilities is rich and varied, with possible effects manifesting on a range of physical scales: from table-top and terrestrial experiments to astrophysical and cosmological observations. As a consequence, making progress requires exploring the possible signatures of new physics in existing measurements, and identifying opportunities to probe these theories with new experimental setups. Here, I collect some of my published work in this direction
Bringing the Military Back In: Military Academy in Latecomer States
No full textThe dissertation complements the existing theories about war and state-making as well as civil-military relations studies by arguing that military academies could nurture disciplined and loyal military elites and transfer knowledge of state-building and modernity critical to state-building in “latecomer” states. The dissertation hypothesizes that military academies facilitate state-building when 1) the existence of strong and committed political centers indoctrinate the military elites with state-building agenda, 2) the academies operate as disciplinary permeation institutions that promote cultures, disciplines, and knowledge that are beneficial to the state-building agenda, and 3) benign relationship between military and civilian elites promote knowledge-sharing and cooperation. The dissertation focuses on the case of Whampoa Military Academy (WMA), founded by the Kuomintang (KMT) in the Republic of China (ROC) in the 1920s. The dissertation traces the career trajectories of selected cohorts of WMA graduates through analyses of biographical data. The dissertation argues that though the WMA elite constituted pockets of disciplined and effective state machinery in the KMT government, the existence of multiple political centers in the KMT regime and the antagonism between WMA graduates and other political forces in the KMT national revolution alliance severely undermined the WMA graduates’ achievements of disciplinary permeation and contribution to state-building. After 1949, the CCP in mainland China and the KMT in Taiwan selectively adopted legacies of the WMA and incorporated militaristic disciplining in different ways in their state-building agendas. The dissertation also provides comparative case studies of the contribution of military academies and officer corps to state-building in Turkey and Japan since the late nineteenth century. The dissertation proposes that studying military academies provides a new perspective to evaluate the influence of war and military on state-building and has important policy implications on the military and politics in the contemporary world
Lyophilized Antibodies As Tools For Efficient T Cell Activation in CAR T Cell Therapy
No full textT cells are a crucial part of the immune system; they maintain homeostasis and fight infections. CAR-T cell therapy is a type of cancer therapy that leverages T cells to fight cancer. This is done by taking T cells from the patient, engineering them to produce a chimeric antigen receptor (CAR), and then reintroducing them into the body. In order for the CAR to be expressed, the T cells must be activated.
T-cell activation involves primary and secondary signaling with the use of two proteins: CD3 and CD28. T cell activation is commonly done using TransAct, a miltenyi biotec product. This product contains CD3 and CD28 monoclonal antibodies. The goal of this experiment is to determine if lyophilized antibodies are an effective method to activate T cells for CAR production.
The Lyostim bag is an alternative method of T cell activation. It is created by lyophilizing CD3 and CD28 antibodies that are suspended in an experimental solvent. The experimental solvents used in this experiment were PBS, Trehalose, and Water. After lyophilized, enriched T cells were added to the bag for activation. The cells were then transduced with 3 different lentiviruses corresponding to different types of cancer.
The Lyostim bag effectively activated all cells. The metrics used to determine cell activation were LDLr expression, cell diameter, viability, and recovery. Looking at CAR% and Total Viable CAR + cells, the Lyostim bags had results greater than or comparable to TransAct.
The Lyostim bag is also a more efficient technology from both an economic and a manufacturing standpoint. To activate 100 million cells, the raw materials of the Lyostim bag cost 2,400.00. From a raw materials standpoint, the Lyostim bag is $2,124.00 cheaper than TransAct. Additionally, in a manufacturing setting the predeposited antibodies of the Lyostim bag create an easy-to-use “grab and go” solution. Overall, the Lyostim bag is a cheap and efficient tool for activating T cells in CAR T cell therapy
How Contractility Kits Can Be Used To Fight Cancer
Get PDFMutations that alter the flexibility of the cell’s cortex can contribute to many diseases, including cancer development and metastasis. This property of cells is not well understood and is the focus of the Robinson Lab at Johns Hopkins University School of Medicine. Their research has led to the discovery of protein condensates known as Contractility Kits (CKs), which has provided insight into how cell cortices respond to mechanical stress and how this flexibility is regulated. The ability to reduce the flexibility of cancer cell cortices has shown promise in developing therapies that prevent metastasis. Currently, the Robinson Lab has two patents for targeted use of the compound 4-HAP, which has shown the ability to increase the concentration of non-muscle myosin II, a CK component that accumulates in the cortex. The ability to make cancer cells more rigid has shown promise in fighting pancreatic and colorectal cancers as well as potentially others.
This research is at the forefront of molecular and cellular biology, and the creation of visuals is catching up. Until now, these processes have been described mostly through static figure illustrations and video clips based on protein concentrations as their structures and behaviors come into focus. However, the association of CKs with the cortex and the behaviors of individual proteins that comprise them are dynamic. What is missing are educational animations of how these proteins sense mechanical force, how they associate and dissociate from the cortex and how this behavior can be modified.
To address this gap in visual material, we created a short 3D animation to describe the proteins that form CKs, as well as what is currently known about the mechanical feedback loops that govern CKs, and finally how small molecular modulators such as 4-HAP can alter the flexibility of cell cortices. This animation will supplement Dr. Robinson’s lecture material and help support his lab’s efforts to get the resources needed for human trials. The video can be found on Dr. Robinson’s website and YouTube, where anyone can access it
LEGIONELLA PNEUMOPHILA MODULATION OF HOST CELLS THROUGH THE VIRULENCE PROTEINS MAVP AND LPG2888
No full textLegionella pneumophila is an intracellular bacterial pathogen that can replicate within human alveolar macrophages, causing pneumonia. One of the mechanisms by which L. pneumophila modulate the host cell is through secreted virulence proteins termed effectors. One such effector designated MavP has been implicated in interacting with the host cell protein PMP70 and functions to recruit peroxisomes to the Legionella membrane-bound replication compartment termed the Legionella-containing vacuole (LCV). MavP has been found to have coevolved with another L. pneumophila effector protein Lpg2888, indicating that the two effectors may be functionally related. The role of Lpg2888 has not yet been elucidated. To define MavP domains that mediate the interaction of MavP with PMP70, the importance of individual MavP domains and putative motifs for recruiting peroxisomes to the LCV were examined. We found that a series of highly conserved polar and charged amino acids in the C-terminus of MavP are important for recruiting PMP70 to the LCV. These results define a novel motif and potential MavP interaction interface that enables MavP to bind PMP70. In parallel, the importance of lpg2888 for L. pneumophila pathogenesis was examined by testing whether deletion of lpg2888 impacts L. pneumophila growth within macrophages.Notably, loss of lpg2888 resulted in significantly reduced L. pneumophila intracellular growth, demonstrating an important role for this gene during infection. The lpg2888 mutant phenotype could be rescued by in trans complementation. These results establish Lpg2888 as a novel L. pneumophila virulence factor, and a foundation for interrogating the roles of MavP and Lpg2888 in disease
EXTRAVASCULAR SPACES ARE THE PRIMARY RESERVOIRS OF ANTIGENIC DIVERSITY IN TRYPANOSOMA BRUCEI INFECTIONS
No full textThe protozoan parasite Trypanosoma brucei evades clearance by the host immune system through antigenic variation of its dense variant surface glycoprotein (VSG) coat, periodically “switching” expression of the VSG using a large genomic repertoire of VSG-encoding genes. Studies of antigenic variation in vivo have focused near exclusively on parasites in the bloodstream using intraperitoneal injections, but recent work has shown that many, if not most, parasites reside in the interstitial spaces of tissues. We sought to explore the dynamics of antigenic variation in extravascular parasite populations using VSG-seq, a high-throughput sequencing approach for profiling VSGs expressed in populations of T. brucei. First, we optimized the most consistent mode of infecting mice by comparing intraperitoneal (IP) and intravenous (IV) injection models, revealing distinct infection dynamics with IP infections likely causing delayed parasite dissemination compared to IV infections. Next, using IV infections, we show that tissues, not the blood, are the primary reservoir of antigenic diversity during T. brucei infections, with more than 75% of VSGs found exclusively within extravascular spaces. We found that this increased diversity is correlated with slower parasite clearance in tissue spaces. These results were additionally reproducible in more natural tsetse bite-initiated infections and across multiple T. brucei cell lines. Together, these data support a model in which the slower immune response in extravascular spaces provides more time to generate the antigenic diversity needed to maintain a chronic infection. Our findings reveal the important role that extravascular spaces can play in pathogen diversification
THE ASSOCIATION BETWEEN FOOD ACCESS AND LIFE EXPECTANCY AT THE CENSUS TRACT-LEVEL IN MARYLAND, 2010-2015
Get PDFBackground
Life expectancy is an aggregated metric often used to indicate the overall health of a population. Low-quality food environments affect dietary decisions and in turn, may impact risk of chronic disease. The association between food access and life expectancy is not fully understood at the census tract-level. Additionally, differences in this association by urbanicity are not clear.
Methods
Using data from the U.S. Small-area Life Expectancy Estimates Project (USALEEP), and the USDA Food Access Research Atlas, we assessed the relationship between food access and other covariates with life expectancy at birth in urban and rural census tracts in Maryland. Low food access is defined as the percent of the population residing outside a 0.5-mile distance from supermarkets in urban areas or a 10-mile distance in rural areas. Crude and adjusted linear regression models were used to estimate the change in life expectancy [95% confidence interval] stratified by urban and rural census tracts.
Results
Census tract-level life expectancy at birth ranged from 62.6 years to 96.1 years, with a mean of 78.7 years. The crude and adjusted models for urban tracts (n=1,096) estimated an increase of 5 percentage points of the population with low food access was associated with a 0.147([0.107, 0.187] crude) and -0.011 ([-0.039, 0.016] adjusted) year change in life expectancy. Among rural census tracts (n=206), the crude and adjusted models estimated a -0.134 ([-0.372, 0.105] crude) and 0.265 ([0.022, 0.508] adjusted) year change in life expectancy was associated with an increase of percentage points of the population with low food access.
Conclusions
Our findings suggest that the association between low food access and life expectancy differs in urban and rural tracts in Maryland. These findings supply important information that must be considered by policy decision makers when assessing and implementing food access interventions in Maryland