Mason Journals (George Mason Univ.)
Not a member yet
3256 research outputs found
Sort by
Exploring the Spatially Dependent Proteome of Ovarian Cancer Tissue
No full textOvarian cancer is a formidable disease often associated with a poor prognosis due to its acquired drug resistance, lack of early detection methods, and high mortality rate. There remains an urgent need to develop new approaches for early diagnosis of ovarian cancer and strategies for reversing drug resistance. We hypothesize that ovarian cancer cells interact with host cells in the tumor microenvironment in a poorly understood mechanism resulting in drug resistance. The tumor microenvironment is very complex, consisting of multiple cell types. In addition, the tissue cells continuously communicate with one another through the secretion and uptake of extracellular vesicles. Recently, we developed a new generation of Laser Capture Microdissection (LCM) to interrogate the tumor microenvironment by isolating specific cells from a heterogeneous tissue section and separating them into distinct homogeneous samples. We used this technology coupled with mass spectrometry (MS) to compare the proteome of extracellular vesicles to the tumor cells subpopulations. In order to explore the function of the differential expressed proteins, we utilized a cell culture model of human ovarian cancer cells as they adapted to cisplatin treatment. Finally, we investigated the spatially dependent relationship between cisplatin therapy and p53 expression. With our newly developed methods in tissue molecular profiling, we were able to highlight the importance of p53 intracellularly and in the interstitial space
PINK-1 containing extracellular vesicles: A new molecular measurement of mitochondrial health and tumor growth
No full textMitophagy is an important survival mechanism to eliminate damaged mitochondria via shuttling to the lysosome for destruction.Tumor hypoxia, chemotherapy, and immunotherapy damage mitochondria. Consequently, mitophagy plays a guardian role in cancer pathogenesis. PINK1 is a mitochondrial membrane sensor to initiate mitophagy. Healthy mitochondria break downPINK1 into fragments by protease cleavage of the full length (FL)-PINK1. In contrast, dysfunctional mitochondria fail to cleave PINK1, thereby accumulating FL-PINK1 triggering mitophagy. We hypothesize PINK1 cleavage status is a novel measure of a cell’s mitochondrial health. We discovered that PINK1, during high mitophagy demand, exports damagedmitochondria within EVs. We found that tumor interstitial fluid(IF) EVs contain different ratios of FL-PINK1 to c-PINK1. We compared ratios of the PINK1 levels in tumor IF vs normal IF and found non-tumor IF contained only c-PINK1 whereas tumor IF contained more FL-PINK1 indicative of tumor oxidative stress. Applying mitophagy-inducing drugs on 4T1 breast cancer cells generated greater levels of FL-PINK1 EV export. EV-associated PINK1 fragmentation is a sensitive indicator of cellular mitochondrial health. It has been shown that the tumor suppressor, p53, locates to PINK1 during mitochondrial stress. We hypothesize that PINK1+ EVs containp53 to support tumor growth. Using purified 4T1 EV populations, PINK1 was immunoprecipitated and probed for CD81, p-p53, and Alix. We found that p-p53 co-located with PINK1+ EVs, therefore the export of tumor suppressors via secretory mitophagy is a novelpro-tumor mechanism. Future work would include developing a rapid PINK1 fragmentation assay and further analysis of p53-PINK1 localization intracellularly
Assessing the performance of Kolmogorov-Arnold Networks in preventing catastrophic forgetting on continual learning tasks
No full textKolmogorov-Arnold Networks (KANs) are a recently introduced alternative to multi layer perceptrons (MLPs), motivated by interpretability and approximation efficiency. Previous work conjectured that KANs may be suitable for continual learning, which is a learning paradigm where data is presented to a learning algorithm sequentially and may change over time. This conjecture was supported by initial results for learning a single-variable target function in a continual learning setting, where the KAN demonstrated negligible forgetting compared to the baseline MLP. Utilizing the pykan package, we investigated the performance of KANs for more difficult continual learning settings than considered in previous work: we evaluated KANs with deeper architectures using more difficult datasets, including multi-variable target functions and Split-MNIST (a standard digit recognition benchmark for continual learning). When learning simple functions with small KANs, we reproduce the previous conclusion that KANs achieve minor forgetting compared to MLPs. However, on Split-MNIST, both KANs and MLPs suffered from catastrophic forgetting
Using data from the EUV-imaging spectrometer to analyze velocities along a magnetic loop at different spectral lines
No full textThe solar system’s star, the sun, goes through 11 year activity cycles. Within the core of the dense sun, energy is created to heat the star through nuclear fusion. This heat travels to outer layers of the sun. Intuitively, the parts of the sun farthest from the core should be the coolest, however; the corona, the outermost layer of the sun, is hotter than its surface (the photosphere), reaching temperatures above 1,000,000 K. The sun’s corona includes entangled magnetic field lines creating loops that are comprised of ionized particles that can jet out into outer space and impact the geomagnetic activity on Earth. Understanding the sun’s magnetic activity, therefore, offers a better understanding of our solar system. To provide observations that can be compared to computer models, we have measured the velocity of the ionized gas in a magnetic loop. We used data from the Hinode satellite’s EUV-imaging spectrometer. For analysis of the Hinode Satellite’s data, the EIS Python Analysis Code was used. A series of points were selected on the loop along one spectral line. After selecting points on the loop, python algorithms were used to collect the velocities at each point. To collect velocities along different temperatures, a Doppler Shift function was applied to correct for differences in the speed of various spectral lines. By looking at the velocity graphs created in the context of expected value ranges, more can be understood on the nature of the sun’s corona. The results of our measurements will be presented
Monitoring Water and Vegetation Change in the Great Green Wall of Africa Using NASA's FLDA Products and Machine Learning
No full textThe Great Green Wall (GGW) is an international effort to grow a vast belt of trees, vegetation, and fertile land across the drylands of the Sahel in Africa. The Great Green Wall (GGW) initiative, launched by the African Union in 2007, aims to combat degraded lands, enhance food security, and build climate change resilliance in the Sahel region. This study leverages multi-dimensional data from FLDAs (Field Level Data Aggregators) to analyze the progress and impact of the GGW from 2007 to 2024. The use of FLDAs’ products has enabled precise, localized data collection and analysis, offering valuable insights into the interactions between vegetation growth and climatic factors. Monitoring key indicators such as evapotranspiration, soil moisture, air temperature, and Normalized Difference Vegetation Index (NDVI) provides a comprehensive assessment of ecological and climatic changes associated with the GGW. Byanalyzing spatio-temporal features, we can better understand the dynamics and patterns of climate and environment changes over time and across different regions. The research underscores the importance of this project in addressing critical challenges within the Africa Water-Energy-Food-Health (AWEFH) Nexus, highlighting its role in the effective management of natural resources. Despite the progress made, a significant gap remains in understanding the long-term socio-economic impacts and the scalability of the GGW across different regions. Additionally, the initiative faces a lack of funding, which threatens its sustainability andexpansion. This study not only contributes to the scientific understanding of large-scale environmental initiatives but also supports policy-making and strategic planning for sustainable development in Africa. Furthermore, it aims to promote awareness and advocate for increased funding to ensure the continued success and expansion of the GGW. Further research is neededto bridge the gap in socio-economic assessments and explore the broader implications of the GGW on local communities and economies
Prevalence and Risk Factor Analysis of Campylobacter in Rural Bangladesh
No full textCampylobacter is the one of the leading bacterial causes of diarrheal disease in the United States. In low and middle-income countries, Campylobacter has been found to be one of the most common diarrheal diseases in children under the age of 5. There are multiple factors that affect transmission, including consuming raw poultry and meat. A large study was undertaken by an international team and data from that study were used for this project. The objective of this study is to describe the strains of Campylobacter carried by humans and animals. 1,927 samples (1,291 human; 635 animal) were screened for Campylobacter jejuni and coli using culture and PCR in Bangladesh, followed by whole genome sequencing (WGS) of isolates. Out of 109 isolates of Campylobacter detected, the majority were C. jejuni at 80 isolates, while the remaining 29 were C. coli. The majority of the 86 chicken stool isolates consisted of C. jejuni, while the 4 pigeon stool isolates contained 2 strains of each C. jejuni and C. coli. The 2 newborn isolates contained solely C. coli, which was also the majority in 7 isolates of the samples from pregnant women, while all 10 isolates from siblings were strains of C. jejuni. The results from this study allow us to understand the differences between Campylobacter carried by humans and animals
Mathematical Modeling and Physics Informed Neural Network approaches for studying the environmental impact of data centers on a county level
No full textLoudoun county in the state of Virginia in the United States is the world’s data center hub, with over 70% of global internet traffic flowing through the county. Rapid expansion of the internet and AI are accelerating data center growth, energy and water use, and emissions, posing a challenge to the UN Sustainable Development Goal (SDG) of net zero emissions by 2050. While estimates of global emissions from data centers exist, this would be the first study to estimate the direct and indirect environmental impact of data centers at the county level. Our study dynamically models the relationship between data center growth, population growth, and increased CO2 emissions using a system of coupled ordinary differential equations. The mathematical model thus accounts for the broader implications of data center concentration, such as its role in stimulating further infrastructure and land development, and assesses the impact on human mortality. Physics Informed Neural Networks (PINNs) are used with input from real-time data in Loudoun County to quantify parameters. Findings identify key causes and impacts of emissions related to data center growth at a local level, and define quantitatively the problem that sustainable energy solutions must address
Signaling storming attack detection and mitigation in open radio access networks
No full textOpen radio access networks (O-RAN) have revolutionized mobile communications by enhancing interoperability and fostering innovation through open interfaces and modular components. However, this flexibility and openness also introduce security vulnerabilities, particularly to signaling storm attacks that can disrupt network operations. Signaling storms occur when devices aggressively attempt to register or attach to the network at a high frequency, potentially orchestrated by an attacker to cause network disruptions in an O-RAN system: such attacks can be particularly detrimental due to the modular and open nature of the architecture. Currently, there is a lack of effective tools to swiftly detect and mitigate signaling storm attacks in O-RAN systems. This project aims to address this gap by developing a software microservice, i.e., xApp within O-RAN’s Near-real-time RAN Intelligent Controller that can detect and mitigate signaling storm attacks in near-real-time (in range of several milliseconds), thereby safeguarding the network's integrity and performance.. The proposed solution involves advanced detection algorithms and mitigation strategies implemented within an xApp framework, enabling real-time response to signaling storms. Key methodologies include the analysis of network traffic patterns and the development of machine learning models to identify abnormal behavior indicative of signaling storm attacks
Using Motion Capture Technology to Learn 6-DoF Vehicle Kinodynamics for Autonomous Navigation on Vertically Challenging Terrain
No full textTraditional autonomous robots limit themselves to flat surfaces and unobstructed navigation trajectories, categorizing their environment into either free space or obstacles, and treating obstacles as untraversable terrain. Recent research has demonstrated that wheeled robots have the potential for off-road autonomous navigation across vertically challenging terrain (rocks, boulders, debris). However, to navigate successfully, autonomous robots need accurate kinodynamic models to compute motion trajectories and predict vehicle-terrain interactions. Most wheeled robots use simplified models such as Ackermann-steering or differential drive which assume that vehicle motion is restricted to a 2D plane and doesn’t account for complex underlying terrain. Machine learning has been used to develop models that can efficiently predict vehicle-terrain dynamics in order to plan trajectories on vertically challenging terrain. Our work aims to improve the accuracy of these models through the use of motion capture technology. First, we set up a 4-camera OptiTrack motion capture system surrounding a 3.1m by 1.3m rock testbed. Next, we equip a 1/10th scale vehicle with numerous infrared markers, allowing its pose to be tracked in real-time with ~1mm accuracy. Finally, by manually driving the vehicle over the rock testbed, we collect a dataset of vehicle-terrain interactions including a terrain elevation map, the tracked robot pose, and corresponding control inputs. This dataset will be used to train a supervised machine learning model to understand vehicle-terrain dynamics (predict future robot poses given the terrain map and control input). Additionally, we developed an equivalent tracked vehicle out of a Traxxas chassis, Azure Kinect depth camera, and NVIDIA Jetson Orin compute module. We then use the motion capture system to collect data on the tracked vehicle, verifying the applicability of our technique across multiple platforms. Overall, our work provides a high-quality dataset to improve navigation on challenging terrain and develops an additional robotic platform to verify the versatility of these navigation techniques across different hardware.
Citations:
Datar, C. Pan, M. Nazeri, i, A. Pokhrel, and X. Xiao, “Terrain-Attentive Learning for Efficient 6-DoF Kinodynamic Modeling on Vertically Challenging Terrain,” in 2024 IEEE International Conference on Intelligent Robots and Systems. (IROS). IEEE, 2024.
Datar, C. Pan, and X. Xiao, “Learning to model and plan for wheeled mobility on vertically challenging terrain,” arXiv preprint arXiv:2306.11611, 2023.
Datar, C. Pan, M. Nazeri, and X. Xiao, “Toward wheeled mobility on vertically challenging terrain: Platforms, datasets, and algorithms,” in 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024
Cross-Generation Reliability Study of NVIDIA Volta and Ampere GPUs on Supercomputers: Similarities and Differences
No full textHigh-performance computing (HPC) operations, like the training of large AI models, hurricane prediction simulations, and climate modeling, rely on the accuracy of GPUs to keep computing costs low and increase power efficiency. By understanding the behavior of Double-Bit Errors (DBEs) in GPUs and examining the progression of GPU reliability across architecture generations, we can gauge how improvements in previous technologies have affected reliability and gain insights into how effective or ineffective changes across generations were.We analyze the Oak Ridge Leadership Computing Facility (OLCF) Summit Supercomputer GPU Snapshots dataset, an extensive dataset taken across two years on the occurrence of double-bit errors in 27,648 Tesla V100 GPUs on the Summit Supercomputer, one of the top 10 supercomputers in the world. Our analysis of the V100 data focuses on several error characteristics, including GPU location within a Summit node, the number of daily errors, and the time between errors. We utilize Pandas and NumPy for data organization and analysis alongside Matplotlib for data visualization to gain better insights into the overall reliability and possible error patterns of the V100 GPU. Furthermore, we compare data visualizations of the V100 GPUs to their successor, the A100 GPUs, to perform a cross-generation study. Preliminary data suggest that location and a history of failures affect future DBEs. Previous studies and initial outputs support that 1) bursty error patterns exist in both generations and 2) V100 is less reliable than the newer A100 GPUs