71446 research outputs found
Sort by
Somatosensory input shapes an S1-ACC circuit that impacts social behaviors in mice
No full textAutism spectrum disorder (ASD) is diagnosed by changes in social behavior and restricted, repetitive behaviors. Additionally, ~95% of autistic individuals exhibit sensory disruptions, including abnormal responses to touch. Multiple ASD models, including Shank3 mutant mice, exhibit touch over-reactivity that is due to abnormal function of peripheral somatosensory neurons innervating the skin. Peripheral somatosensory neuron dysfunction during development causes disruptions in the region of primary somatosensory cortex that processes touch to the body (S1TR), and social impairments in adult mice. However, the mechanisms through which peripheral somatosensory neuron dysfunction leads to social behavior deficits in mouse models for ASD are unknown.
We hypothesized that peripheral somatosensory neuron dysfunction alters S1TR functions, which disrupts S1TR long-range connections with circuits that modulate social behaviors. S1 integrates somatosensory information and can influence complex behaviors via long-range projections. Yet, most studies of S1 in mice have focused on the subregion of S1 that receives whisker inputs (S1barrel), while little is known about S1TR long-range connectivity and how it may differ from that of S1barrel. We found that S1TR and S1barrel long-range connectivity throughout the brain differ dramatically, including a unique S1TR projection that targets the rostral part of ACC (rACC), which does not receive inputs from S1barrel. Using in vivo multiunit electrode recordings, we observed robust and heterogeneous responses to touch stimuli in rACC neurons and found that the S1TR projection to rACC is required for these responses.
As ACC is known to be important for shaping social behaviors, we hypothesized that peripheral somatosensory neuron dysfunction alters S1TR-rACC circuitry, which alters touch stimulus encoding in rACC and disrupts social behaviors. We found that S1TR-ACC anatomical connectivity is decreased in Shank3 mutant mice. We also found that S1TR-rACC neurons may be dysfunctional in Shank3 mutants and that rACC neuron responses to light touch stimuli are decreased and less reliable in mice with germline or selective loss of Shank3 in peripheral somatosensory neurons. Lastly, we found that optogenetic activation of S1TR-ACC neurons promotes social behavior in control mice, but S1TR-ACC modulation of social behaviors is attenuated in Shank3 mutants, likely due to altered anatomical and functional properties of this S1TR-rACC circuit. Thus, our results identify how altered tactile inputs, beginning with peripheral somatosensory neuron dysfunction, disrupt the anatomy and function of an S1TR-rACC circuit that then contributes to social behavior abnormalities in mouse models for ASD.Neuroscienc
Functional & Taxonomic Identification of Microbial Inflammation in COVID-19 & Parkinson’s Disease: A Metagenomic Analysis
Get PDFPast metagenomic research links disruptions in the gut microbiome to various inflammatory and neurological conditions, including Parkinson’s disease (PD). Recent studies show that PD and SARS-CoV-2 (COVID-19) infections are associated with decreased microbial diversity, altered renin-angiotensin system (RAS) signaling, and increased oxidative stress. This thesis hypothesized a shared dysbiotic signature between these conditions and sought to identify overlaps relative to neurologically healthy controls (NHC). A metagenomic analysis of publicly available datasets from Wallen et al., 2022 (PRJNA834801), and Nguyen et al., 2023 (PRJNA976404) was performed to characterize microbial taxonomy (MetaPhlAn2), functional pathways (HUMAnN2), and to identify statistical biomarkers (QIIME2, MaAsLin2). The goal was to compare PD and COVID-19 metagenomes. Contrary to the initial hypothesis, results revealed distinct, independent microbial signatures. The COVID-19 group showed significant taxonomic imbalance, with increased alpha diversity, lower levels of beneficial bacteria, and higher levels of opportunistic bacteria. In contrast, the PD group's microbiome was similar to that of the NHC group, both taxonomically and functionally. Despite notable taxonomic shifts in COVID-19, functional diversity remained stable across all groups, indicating high functional redundancy. Although no standard inflammatory profile was identified, this study provides a robust workflow for the computational analysis of metagenomic data.Extension Studie
Investigating Differential Expression on Birds from Mongolia Based on Aridity
Get PDFThe Mongolian climate has a division between northern and southern locations based on arid environmental variables: mean annual temperature and precipitation. This study used RNA-seq data from three different species and tissue types to test for differential gene expression. There were nine combinations of species and tissue types and all but one had at least one differentially expressed gene. One combination had significant differences in expression based on northern and southern locations, the Daurian Redstart (Phoenicurus auroreus) and muscle tissue with 800 differentially expressed genes at p.05 significance. The findings demonstrate that there is differential expression based on aridity for at least one species and tissue from Mongolia.Extension Studie
Metal–Organic Frameworks for Permanent Microporosity in Aqueous Media
No full textAqueous media with high gas solubility are critical to the development of many emerging biomedical and energy technologies. From a biological standpoint, most physiological processes depend on cellular interaction of gases with water, while the design and synthesis of sustainable energy materials often require efficient gas-liquid mass transfer in an aqueous medium. However, water lacks the ability to store and transport significant amounts of gas due to its high energetic penalty against cavity formation required to solubilize gas molecules. To overcome this limitation, herein, we establish a generalizable method towards instilling permanent microporosity in aqueous media for increased gas carrying capacities. Specifically, we utilize metal–organic frameworks (MOFs) to instill stable, hydrophobic pores in aqueous media that can readily adsorb gas while excluding water from their pores. This novel approach towards creating aqueous gas carriers allows for not only unprecedented gas solubility in water but also provides a model platform upon which the interfacial effects between water and the micropore can be investigated.
Chapter One introduces gas solubilization in water, and the associated thermodynamic challenges. The concept of instilling permanent free volume towards increased gas solubility is introduced in the context of porous liquids, and the limits of its steric exclusion approach in aqueous systems discussed. The grounds for establishing thermodynamic exclusion of water molecules from hydrophobic MOF micropores are proposed instead to establish permanent microporosity in water.
Chapter Two outlines the design and synthesis of aqueous gas carriers via stably dispersing hydrophobic microporous solids – termed “microporous water” – and the development of in situ analytical methods for directly probing the gas solubility and release of these microporous water systems. Through this study, we report that these microporous water systems can concentrate gases to densities magnitudes higher than what is possible for other aqueous gas carriers, which has exciting implications for biomedical and energy applications.
Chapters Three and Four investigate the fundamental factors that govern water intrusion into hydrophobic micropores through a 1-dimensional (1D) pore channel MOF. We find that pore size, in addition to ligand hydrophobicity, plays a critical role in determining water intrusion into 1D micropores and establish design principles towards de novo synthesis of 1D channel MOFs with microporous water behavior. In addition to structural factors that influence water intrusion, we also explore the effects of pore chemical environment on water intrusion behavior, in which the ligands of a fully water-intruded MOF are systematically functionalized to understand the effects of varying degrees of hydrophobicity and steric bulk.
Chapter Five extends the microporous water concept towards microporous hydrogels for applications in controlled gas delivery. Through leveraging the surface hydrophobicity of aqueous MOF dispersions, we can achieve colloidal MOF hydrogels without the addition of extensive polymer matrices that may infiltrate the pores at the detriment of gas capacity. This proof-of-concept system opens a new avenue for incorporating dry porosity in hydrogels, enabling the sustained delivery of gaseous species previously considered elusive within a water-spanned matrix.Chemical Physic
Designing for Decentralized Finance through Differentiable Optimization, and a Study of Bayesian Optimization
Get PDFDecentralized finance (DeFi) is the concept of building financial infrastructures without relying on centralized intermediaries. A notable development in DeFi is the creation of decentralized exchanges (DEXs), which operate as smart contracts on a blockchain. Due to the high cost of on-chain operations, automated market makers (AMMs) such as Uniswap v3 have emerged as the prevailing model of liquidity provision on DEXs. Two closely related research questions arise in the DeFi space: (1) What are the optimal strategies of liquidity providers given an AMM design such as Uniswap v3? (2) How should the design of AMMs be optimized to achieve objectives such as profit maximization? This thesis addresses these two central research questions using computational methods, in particular, through differentiable optimization.
Chapters 2 and 3 study the optimal strategies of liquidity providers (LPs) in Uniswap v3. In both chapters' formulations, the expected utility of an LP is differentiable with respect to its liquidity allocation under any exogenous price sequence, enabling differentiable optimization of LP strategies. With the formulation of a convex stochastic optimization problem that can be solved in a differentiable manner, Chapter 2 explores optimal static LP strategies in economic settings with varying factors such as an LP's belief about price dynamics, risk aversion, and for different specifications of the Uniswap v3 liquidity pool. Understanding LP strategies also leads to insights into the design of Uniswap v3 liquidity pools. Under a similar optimization framework, Chapter 3 extends from static LP strategies to dynamic LP strategies, specifically LP strategies that reallocate liquidity whenever the price movement reaches a certain threshold. These proposed dynamic strategies—particularly context-dependent variants modeled by a neural network, which adapt the shape of liquidity allocation to contextual information such as price and moving average of non-arbitrage trade volume at the time of reallocation—are shown to lead to significant gains compared to static LP strategies.
Taking a broader perspective on AMM design, Chapter 4 optimizes market-making mechanisms for a single trade in settings with multiple traded goods, seeking market maker profit maximization under adverse selection. Conjectures of optimal mechanisms are generated using tools of differentiable economics, which uses differentiable optimization for economic design. To prove the optimality of proposed mechanisms, a duality theorem is established between the market-making mechanism design problem and an optimal transport problem. This approach of combining differentiable economics with theoretical analysis is used to develop a parameterized class of optimal market-making mechanisms. These results also establish that, in some cases, the optimal market maker across multiple goods must use complex bundling. Additional conjectures about the structure of optimal mechanisms are presented, and an empirical optimality bound is established for some conjectures by approximately solving the dual with linear programming.
The second part of this thesis studies transfer learning of the Gaussian process (GP) prior in Bayesian optimization (BO), a widely used black-box function optimization method. Previous GP-based transfer learning methods for BO are limited to utilizing historical data collected from black-box functions with the same domain as the new black-box function to be optimized. The proposed method, model pre-training on heterogeneous domains (MPHD), employs a neural network that maps from domain-specific contextual information to specifications of hierarchical GPs for a given domain. As a result, MPHD is able to transfer knowledge across heterogeneous domains such as hyperparameter-tuning for different machine learning models. It is shown through theoretical analysis and empirical results that MPHD is a practical transfer learning method for BO, with demonstrations of competitive performance on challenging real-world hyperparameter-tuning tasks.Engineering and Applied Sciences - Computer Scienc
Discovery of a general mechanism for bacterial cell envelope polymer acylation
No full textBacteria frequently decorate their cell envelope polymers with acyl groups that regulate physiology, enhance virulence, and contribute to antibiotic resistance. How cells tackle the challenge of moving activated acyl groups from the cytoplasm—where they are made—onto extracytoplasmic polymers has been a longstanding question. In the work described in this thesis, I uncover a widespread strategy for bacterial cell envelope polymer acylation in which a membrane-bound O-acyltransferase (MBOAT) protein transfers acyl groups from intracellular thioester donors to the side-chain hydroxyl group of an extracytoplasmic tyrosine residue. The acylated tyrosine then serves as a donor for a separate transferase responsible for moving acyl groups to their next destination, usually the cell envelope polymer itself. In the pathway for D-alanylation of lipoteichoic acids, which is the primary focus of this thesis, the key extracytoplasmic tyrosine is located within a highly conserved six-amino acid motif at the C-terminus of a small membrane protein called DltX that holds the MBOAT protein and the other transferase together in a tripartite complex. For other pathways found in diverse bacteria and some archaea, the tyrosine is present in a similar six-amino motif at the C-terminus of the MBOAT protein itself. This work establishes that the function of the vast majority of bacterial MBOAT proteins is to produce acyl-tyrosine intermediates critical for cell envelope polymer modification, setting the stage for function-informed development of inhibitors that target these proteins.
In Chapter 1 of this thesis, I introduce bacterial cell envelope synthesis and modification as an important target for antimicrobial therapeutics. I also describe the history of the study of bacterial MBOAT-based cell envelope polymer acylation systems. In Chapter 2, I investigate the identity of the putative unknown intermediate in the long-studied lipoteichoic acid D-alanylation pathway, and I explain how I developed the hypothesis that a tyrosine-containing motif at the C-terminus of the small protein DltX is the key “missing piece” to that pathway’s mechanistic puzzle. I go on to describe how the work on DltX led me to propose that the specific mechanism I discovered is widespread across different polymer acylation pathways in diverse bacteria. Next, in Chapter 3, I collaborate to use in vitro biochemistry and structural biology approaches to provide solid experimental evidence for my proposed mechanism, again focusing primarily on the protein machinery responsible for lipoteichoic acid D-alanylation. Finally, I conclude in Chapter 4 with a discussion of interesting avenues for future exploration related to both the MBOAT-based acyl transfer pathways I explored in my thesis work and also acyl transfer pathways that use a different class of membrane-bound acyltransferases which I became interested in over the course of my studies.Chemical Biolog
Spatio-Temporal Methods for Causal Inference in Quasi-Experimental Studies: Applications to Environmental Health
Get PDFModern environmental health studies often rely on quasi-experimental designs, where policies or external shocks induce localized changes in exposure across geographic regions. When comprehensive panel data are available before and after an intervention, one can, in principle, leverage the observed data to reconstruct counterfactual trends. Yet rare outcomes (e.g., low counts of a disease) and unmeasured confounding that evolves dynamically across regions and periods can undermine standard approaches like difference-in-differences or synthetic control methods. This dissertation develops a unified suite of Bayesian spatio-temporal methods, spatio-temporal matrix completion and Gaussian process models, that explicitly borrow strength across units and time points to stabilize inference and quantify uncertainty. Through extensive simulations and real-data applications, we demonstrate how these methods generalize popular causal tools, yield interpretable weighting schemes, and provide practical guidance on model implementation for environmental health research.
In Chapter 1, we introduce Bayesian spatio-temporal matrix completion models tailored for rare count outcomes in quasi-experimental panel data through an application examining the impacts of traffic-related air pollution (TRAP) on childhood hematologic cancers. Although some pollutants emitted in vehicle exhaust, such as benzene, are known to cause leukemia in adults with high exposure levels, less is known about the relationship between TRAP and childhood hematologic cancer. In the 1990s, the US EPA enacted the reformulated gasoline program in select areas of the US, which drastically reduced ambient TRAP in affected areas. This created an ideal quasi-experiment to study the effects of TRAP on childhood hematologic cancers. However, existing methods for quasi-experimental analyses can perform poorly when outcomes are rare and unstable, as with childhood cancer incidence. We develop Bayesian spatio-temporal matrix completion methods to conduct causal inference in quasi-experimental settings with rare outcomes. Selective information sharing across space and time enables stable estimation, and the Bayesian approach facilitates uncertainty quantification. We evaluate the methods through simulations and apply them to estimate the causal effects of TRAP on childhood leukemia and lymphoma.
In Chapter 2, we expand on Chapter 1 to investigate the potential heterogeneous impacts of the reformulated gasoline program on the incidence of CYA lymphoma across disease type and demographic strata. We employ recently-proposed Bayesian causal Gaussian process (GP) models, applied to population cancer registry data, to estimate effects of the program on CYA lymphoma incidence across strata defined by cancer type, sex, race, Hispanic ethnicity, and age group. Our analytic framework allows for stable estimation of stratum-specific effects via data-driven information sharing across space, time, and strata. Effects are reported on both the absolute and relative scales. We find evidence that the largest program-attributable reductions in lymphoma incidence rates occurred for Hodgkin lymphoma, and among individuals who are male, white, and/or aged 20-29. The finding of larger reductions in Hodgkin lymphoma is notable since prior TRAP studies have primarily focused on non-Hodgkin lymphoma.
In Chapter 3, we delve deeper into Gaussian process approaches for quasi-experiments, addressing diverse confounding structures that may not be fully accommodated by the model presented in Chapter 2. Estimating causal effects in quasi-experiments with spatio-temporal panel data often requires adjusting for unmeasured confounding that varies across space and time. Gaussian processes offer a flexible, nonparametric modeling approach that can account for such complex dependencies through carefully chosen covariance kernels. In this paper, we provide a practical and interpretable framework for applying GPs to causal inference in panel data settings. We demonstrate how GPs generalize popular methods such as synthetic control and vertical regression, and we show that the GP posterior mean can be represented as a weighted average of observed outcomes, where the weights reflect spatial and temporal similarity. To support applied use, we explore how different kernel choices impact both estimation performance and interpretability, offering guidance for selecting between separable and nonseparable kernels. Through simulations and application to Hurricane Katrina mortality data, we illustrate how GP models can be used to estimate counterfactual outcomes and quantify treatment effects. All code and materials are made publicly available to support reproducibility and encourage adoption. Our results suggest that GPs are a promising and interpretable tool for addressing unmeasured spatio-temporal confounding in quasi-experimental studies.Biostatistic
Building Beyond the Standard Model: Tools from Cosmology to Particle Theory
Get PDFThe current landscape of high-energy theory and cosmology suffers from an abundance of theoretical models coupled with a lack of immediate new data to constrain them. This thesis addresses several open questions in early-universe cosmology and beyond-the-Standard-Model (BSM) particle physics, specifically focusing on inflation, dark matter, axions and axion strings, and baryogenesis. Rather than relying solely on forthcoming experimental results, I argue for refining the BSM model space through theoretical consistency checks and cross-examination against multiple existing datasets. For instance, inflationary models often encounter severe naturalness (or ) problems; applying effective field theory techniques and symmetry considerations helps exclude models that fail these theoretical standards. Similarly, axion models naturally predict distinctive topological defects, such as axion strings, allowing their observational signatures (or lack thereof) to impose valuable constraints. On the observational side, I emphasize that jointly analyzing the Hubble and large-scale structure (LSS) tensions can significantly constrain dark-sector theories. Additionally, I introduce a novel observational probe, analyzing the effect of early universe inhomogeneities generated prior to Big Bang Nucleosynthesis (BBN) on predicted deuterium abundances. This probe rules out baryogenesis scenarios that produce excessive inhomogeneities which are not fully erased by diffusion, and it can potentially constrain regions of parameter space in prominent high energy scale models such as electroweak baryogenesis (EWBG). Together, these theoretical and observational approaches offer robust methods for refining the BSM landscape and demonstrate the critical role particle theorists can play in interpreting cosmological data, even in the absence of new experimental results.Physic
Ambidirectional Liquid Crystal Elastomers: Opposite Deformations Designed Across Length Scales
No full textlEngineering and Applied Sciences - Engineering Science
Incentive Design in the Machine Learning Age
Get PDFThis dissertation investigates the design of incentives in multi-agent systems where traditional assumptions -- such as fully rational agents and omniscient principals -- do not hold. As real-world systems increasingly involve learning-based decision-makers, either human or algorithmic, this work explores how learning alters the landscape of incentive design. The dissertation is organized into three parts.
The first part focuses on incentive design by learning principals, specifically in information and mechanism design settings. For information design, this part introduces novel algorithms that allow a principal to learn an agent's non-Bayesian belief updating process, such as a subjective prior or cognitive bias, via strategic interaction with the agent.
For mechanism design, this part examines how a coordinator can learn to compute Bayes correlated equilibria in non-truthful auctions using limited samples of agent types.
The second part studies incentive design for learning agents, who are modeled as boundedly rational learners rather than best responders. This part first presents, for a general class of principal-agent problems, a reduction from no-regret learning agents to approximately best-responding agents, enabling a precise analysis of the principal's performance. It then characterizes the convergence properties of multi-agent learning in first-price auction games, identifying when convergence to equilibrium is possible.
The third part explores incentive issues in deployed machine learning systems, with a case study on recommender systems. It demonstrates that the strategic behaviors by content creators can exacerbate polarization, even under diversity-promoting algorithms, and proposes alternative algorithmic designs that mitigate these effects.
Collectively, this dissertation lays foundational insights for designing systems that are robust to the incentives of learning-based, data-driven participants.Engineering and Applied Sciences - Computer Scienc