22689 research outputs found
Sort by
Developmental neurotoxicity assessments of neurotoxicants in CRISPR/Cas9-engineered brain organoids
No full textAlthough compared with 2D cells, animal models are able to provide more advanced physiological systems, which benefits the toxicological studies in resembling physiological characteristics, it is always debatable to which extent this resemblance could be applied to humans. This leads to an increasing need for a new model to bridge the gap between these two traditional models. The emerging three-dimensional cell culture system (3D organoids) offers a potential solution to address this issue. Mostly derived from human donor cells, 3D organoid models are considered more human related, and the 3D structure allows them to move away from single cell type composition and enables the appearance of cellular communication, which is of significant importance in the study of development neurotoxicity. The 3D brain organoids developed at the Center for Alternatives to Animal Testing (CAAT) model major neural development processes including synaptogenesis, oligodendrocytes, and subsequent myelination, providing an efficient platform to assess the chemical induced developmental neurotoxicity. By using CRISPR/Cas9 gene editing technology, we now introduced two fluorescent proteins, green fluorescent protein (GFP) to tag a protein expressed in oligodendrocytes (proteolipid protein 1 (PLP1)), and blue fluorescent protein (BFP) to tag a pre-synaptic protein (synaptophysin (SYP)), into the human induced pluripotent stem cell (hiPSC) line and further differentiate the double reporter line into brain organoids. As a resulting functionality, validation as well as the development of neurotoxicity assessment of potential neurotoxicants, we then exposed the CRISPR/Cas9-engineered brain organoid model to four chemicals (lysophosphatidylcholine (LPC), paroxetine, terbutaline, and dibutyl phthalate) and demonstrated their perturbance of synaptogenesis, oligodendrogenesis, and astroglia migration. This study showed the CRISPR/Cas9-engineered PLP1-GFP/SYP-BFP brain organoid as a potential tool to accelerate (developmental) neurotoxicity assessment by combination with high-content imaging. The 3D brain organoid is considered a closer reflection of the human brain and therefore more human-relevant model than traditional 2D cultures. Furthermore, it promises to translate findings to humans, providing means to identify personalized medication and better understandings of the respective diseases
THE DEVELOPMENT OF CHIMERIC ANTIGEN RECEPTOR (CAR) B-CELL THERAPY AND 3D TISSUE RECONSTRUCTION OF PYMT MOUSE MAMMARY TISSUE
No full textCancer is currently the number one cause of death worldwide, and the cancer statistics from Siegel et al. predicted the year of 2023 to have approximately two million new cancer incidents. In recent years, immunotherapy has become a promising field in fighting cancer. In this thesis work, I first exploited the potential of chimeric antigen receptor (CAR)-B therapy for cancer treatment.
B cells are often known as helper cells to other immune cells to fight diseases and infections. However, after the recognition of tumor cells, B cells can also perform direct cancer killing through granzyme B secretion as well as FasL and TRAIL-induced apoptosis pathways. As the potential of CAR-B cells for cancer therapy is unclear, I established CAR-B cells by referencing the second generation of CAR-T cells. My result demonstrates the engineered CAR-B cell has an enhanced capability in eliminating ovarian cancer cells (OVCAR3) when compared to the regular B cells. This result suggests that CAR-B cells can potentially be a new category of immunotherapy in cancer treatment.
In addition, to better understand breast cancer progression in situ in 3D, I reconstructed large volumes of the PyMT mouse mammary models at different progression stages with serial sectioned H&E-stained tissues. I established a deep learning model that can perform automated labeling on tissue sections for 13 different classes. This reconstructed and fully labeled tumor volume can allow us to comprehensively examine tumor progression
Examining the association between carbohydrate metabolism and pancreatic cancer risk using Mendelian Randomization
No full textPancreatic cancer has a high mortality rate with a 5-year survival of only 11.5%, and age-adjusted incident cases are slowly rising. This has led to increased public health concern and a push to identify novel risk factors. Mendelian randomization is a useful tool to investigate specific exposure-outcome relationships and can identify genetic variants that may elevate the risk of disease.
Previous Mendelian randomization studies focusing on pancreatic cancer have investigated type II diabetes, obesity, elevated blood glucose, and other related exposures. There are few that focus on dietary aspects and metabolism, and those that do typically focus on dietary fats. Our study examines genetic variants associated with three metabolites (1,5-anhydroglucitol [1,5-AG], glucose-6-phosphate [G6P], and ribose-5-phosphate [R5P]). These metabolites play a role in carbohydrate metabolism. We hypothesize that genetic variants associated with these metabolites, and therefore changes in levels of carbohydrate metabolism, will be associated with pancreatic cancer risk. Our null hypothesis states that variants (SNPs) associated with carbohydrate metabolism do not affect the risk of pancreatic cancer. Our study investigates seven SNPs associated with 1,5-AG, six SNPs associated with G6P, and five SNPs associated with R5P. Pancreatic cancer genome-wide association summary data was obtained from PanC4, PanScan I and II, and meta-analysis data of PanC4+PanScan I, II, and III. A total of nine Mendelian randomization analyses were completed for each metabolite and each dataset, using five methods (inverse variance weighted, MR Egger, weighted median, simple mode-based, and weighted mode-based). Sensitivity analyses were conducted to examine the validity of our genetic instruments.
Overall, our results did not support a strong association between genetic variants associated with these carbohydrate metabolism phenotypes and pancreatic cancer risk. However, when we examined the PanC4 data alone as well as the meta-analysis data, 1,5-AG showed significant associations using the inverse variance weighted and weighted median methods. We also observed a significant association for G6P and pancreatic cancer risk in the PanScan I/II dataset using the inverse variance weighted method. Though these results indicate that future research into these exposures may be warranted, variance in results across different Mendelian randomization methods creates uncertainty in our findings
Globally Optimal Matching of Astronomy Catalogs Using Mixed Integer Quadratically Constrained Programming and Constrained Clustering
No full textWe propose two new methods for probabalistic cross-matching of astronomy catalogs. Our first method builds on previous work using mixed integer programming by introducing quadratic constraints to shrink the problem by multiple orders of magnitude. Our second method uses constrained k-means clustering to provide approximate solutions for the cross-matching problem. We also provide a method to use the approximate solution to speed up the exact method. We empirically show that our new mixed integer program with quadratic constraints is able to be set up and solved much faster than the previous large linear formulation. We also show empirically that our constrained clustering provides near optimal solutions in a fraction of the runtime of previous methods. We base our findings on simulated catalogs and real-world catalogs from the Hubble Space Telescope. This thesis is accompanied by publicly-available software to demonstrate both algorithms
An Ensemble Approach to Full Resolution Mammogram Classification
No full textBreast cancer remains the most commonly diagnosed cancer in women in the United States, and while treatment advancements have reduced the death rate, the diagnostic process remains complex and expensive and requires significant human expertise. Mammograms remain a vital part of that diagnostic process and can provide an early indication of the presence of a problematic lesion within the breast.
This thesis employs the Curated Breast Imaging Subset of the Digital Database for Screening Mammography (CBIS-DDSM) dataset, the InBreast dataset for testing and a modified version of the VGG16 convolutional neural network to detect and classify suspected benign or malignant lesions within the breast mammogram. The modified VGG16 network accepts images of the mammogram without image size reduction or down-sampling so as to avoid image degradation. In addition, a neural network ensembling methodolgy is employed to explore improvement of the overall classification performance of the design
MEASUREMENT AND MITIGATION OF MICROBIAL CONTAMINATION ON VARIOUS PERSONAL PROTECTIVE EQUIPMENT
No full textRecent global health crisis has called for methods to measure and mitigate microbial contamination on
various personal protective equipment. Particularly, this document focu sed on: the implementation of modified
AATCCAATCC-100 with qRT qRT-PCR assisted absolute and relative quantification for quantifiable tracing of both
antimicrobial and microbial behavioral properties; the comparison of pipette tip repurposing efficiencies among
lab detergent, ozone, and CAP; and the prospective application of vapor hydrogen peroxide, ozone, and CAP for
mask repurposing.
Log reductions from modified AATCC
- 100 were compared to identify time dependent antimicrobial were compared to identify time dependent antimicrobial
properties from silver ion containi containing wound dressing samples. A ntimicrobial properties of wound dressing
samples diminished as incubation days are increased for both PCR and cell viability assay, while d ata from qRT qRT- PCR generally produced lower standard deviation than that of culture method s, hence shown to be more precise.
Complementary parallel analysis of samples using both methods better characterized antimicrobial properties of
the tested samples.
A p
arallel analysis using classical methods alongside the application of relative quantifi cation displayed
changes in expression of virulence related genes. Although molecular assays targeting specific virulence activities
are needed to verify the change in activities, relative quantification efficiently provided insight into changechanges specific t o virulence in model organisms.
A c
ontamination evaluation protocol were outlined t o evaluate the efficacies of the following
repurposing methods: washing wit h a common laboratory detergent, exposure of ozone vapor, and CAP. Efficacy
was determined by turn over ratio and log reduction in detectable genomic material of the contaminated products
via re real -time quantitative PCR (qPCR). Ozone at 14400 PPM * minute is fully optimized while CAP shows
promising potential post optimization.
The application of ozone, hydrogen peroxide, and CAP is further explored for mask repurposing.
Although further experimentation with BFE is needed, minimal change in physical properties of post post-repurposed
masks showed promising potential as non non-destructive repurposing methods
The linear stability of weakly charged and slowly rotating Kerr Newman family of charged black holes
No full textThe Einstein-Maxwell system describes the interaction of gravity and electromagnetism. In Einstein-Maxwell system, Einstein's field equation describes the relation between the geometry of the spacetime and the energy momentum of an electromagnetic field, and Maxwell's equation describes how electromagnetic waves propagate in the spacetime. Among the most interesting solutions to Einstein-Maxwell equations are the families of black hole solutions. The Kerr-Newman family of solutions describes stationary, charged, and rotating black holes.
In this thesis, we prove the linear stability of weakly charged and slowly rotating Kerr-Newman black holes under coupled gravitational and electromagnetic perturbations. We show that the solutions to the linearized Einstein-Maxwell equations decay at an inverse polynomial rate to a linearized Kerr-Newman solution plus a pure gauge term. This work builds on the framework developed in \cite{HHV21} for the study of the Einstein vacuum equations. We work in the generalized wave map and Lorenz gauge. The proof involves the analysis of the resolvent of the Fourier transformed linearized Einstein-Maxwell operator on asymptotically flat spaces, which relies on recent advances in microlocal analysis and non-elliptic Fredholm theory developed in \cite{Vas13}. The most delicate part of the proof is the description of the resolvent at low frequencies
EFFECTIVE ARCHITECTURES AND TRANSFER LEARNING APPROACHES FOR COMPUTER VISION APPLICATIONS
No full textTo train a deep neural network for a specific application, two crucial factors are the network architecture and the data used for training. The architecture of the network determines the types of features that the model can learn to address the problem at hand while transfer learning approaches address the data aspect by determining how effectively a model can generalize to new data and what initial weights are optimal for training the model. Although deep learning has made significant strides in computer vision, there are still many applications where state-of-the-art solutions are not entirely effective. In this thesis, we concentrate on various vision-based issues in medical image analysis and image/video enhancement domains and propose novel architectures and transfer learning techniques to solve them.
Medical imaging presents significant challenges, including limited data availability, specialized tasks, absence of pre-trained models specific to the medical domain, and the fundamental differences between medical and natural images. To address these challenges, we propose three novel architectures tailored for medical imaging tasks: the first focuses on accurate segmentation of small anatomy and sharp boundaries, the second is designed to extract long-range dependencies, and the third is optimized for faster inference to support point-of-care applications. In addition, we introduce three transfer learning approaches specifically designed for medical imaging. These include: a medical pre-training strategy and initialization weights to support radiology-based downstream tasks, a novel test-time adaptation method for use in clinical settings, and a synthetic data generation technique to overcome the challenge of limited data availability. Our efforts also extend to addressing challenges related to image and video enhancement. Specifically, we have developed solutions, including: a video architecture optimized for real-time enhancement of live streams and video conferences, an all-in-one network designed to remove adverse weather effects from images, and an interactive pipeline for image enhancement
UNDERSTANDING THE CONSEQUENCES OF POLYGENIC ARCHITECTURES ON COMPLEX TRAITS AND DISEASE
No full textA fundamental goal of human genetics is to understand the genetic architecture of a trait or disease, a conceptual framework that relates genotype to phenotype. Genetic architectures include considerations of the frequency and effect size of the underlying trait- or disease-associated genetic variation that predisposes phenotypic variation. Unlike Mendelian forms, the genetic architectures of complex traits and disease are polygenic and consist of many associated common variants of small effect spread across the genome. Most of the discovered variation exists in noncoding regulatory DNA. The work presented in this dissertation addresses challenges arising from the study of polygenic architectures. In Chapter 2, I use rare variation to address the challenge of mapping relevant genes to common trait-associated SNPs. I present work that leverages rare alleles existing in a large, outbred population to study variation in the estimated copies of the mitochondrial genome, termed mitochondrial DNA copy number (mtDNA-CN), a complex, polygenic trait that is a core feature of Mendelian mtDNA depletion syndromes but also associated with common aging-related disease. We apply single variant association testing and gene-based burden testing of rare alleles found within 415,422 exomes to study mtDNA-CN. Genes with an excess of rare variant signal are enriched in core pathways relevant for mitochondrial biology. Rare variants also delineate an ancestral haplotype associated with increased mtDNA-CN. In Chapter 3, I use common variation to addresses the challenge of polygenicity. We employ gene expression in a general population as a proxy phenotype to study the collective downstream effects of the highly polygenic common variant architecture of schizophrenia (SCZ). We find several genes whose expression is associated with the combined additive component of polygenic risk, but note that some loci, such as those within the major histocompatibility complex, are seemingly devoid of trans-effects from the SCZ architecture, while others harbor no SCZ risk alleles that affect expression in-cis. We hypothesize the phenotypes associated with the latter, whose expression we demonstrate is driven only by coalescing SCZ trans-eQTL effects, represent bystander phenotypes for SCZ, a consequence of the gene regulatory networks within the SCZ genetic architecture
LEAD CHALCOGENIDE QUANTUM DOT ASSEMBLY AND ATTACHMENT ON FLUID INTERFACES
No full textThe formation of tiles composed of quantum dots is thought to constitute a new class of self- assembled nanostructured material. Understanding the dynamic physicochemical processes that govern the assembly at a functionalized fluid interface is crucial. We aim to seek design guidance for future advances in improving the assembly result of nanostructured nanocrystals (NCs). What is expected is that a functionalized liquid interface potentially provides more control over the behavior of the nanocrystals (NCs), which increases the complexity of the assembly process. To investigate this self-assembly process, we investigate the insight from the fundamental molecular-level interactions, and we use Molecular Dynamics (MD) simulations to investigate the process. The model system studied here was composed of lead chalcogenide NCs, covered with lead oleate molecules (ligands), and assembled on a monolayer (ML) composed of amphiphile (DPPC) molecules. The impact of ML density and structure of amphiphile molecules are extracted as two main factors of interest.
The simulations aimed to reveal the role of the nature of the monolayer interface on NCs assembly process. Before testing the impact of monolayer parameters on the self-assembly of the NCs, we used density functional theory to confirm that the energy barrier for ligand dissociation from the surface of the NC was high and, as such, are unlikely to detach readily. We studied the degree of NC in-plane orientation and out-of-plane tilt degree during the process as two means of evaluating the NCs’ alignment. We generated simulations focusing on the impact of these two factors on self- assembly that will bring us practical insights into how ML density and the chain length of ML molecules affect the self-assembly performance. We uncover a trend of the analytical attachment among various NCs assembly processes that reflects the contributions that ML density and length of alkyl chain made to the formation of defects in dimer alignments. The simulations and experiments presented in this study provide concrete design rules for the assembly of NCs for further research on superlattice formation