12023 research outputs found
Sort by
Imaging Cell Lineage with a Synthetic Digital Recording System
Get PDFIn multicellular organisms, the lineage history and spatial organization of cells both play pivotal roles in cell fate determination during development, homeostasis, and disease. Investigating lineage relationships alongside cell state and space would provide a fundamental understanding of these biological processes. Current lineage tracking approaches rely on the progressive accumulation of either naturally-occurring somatic mutations or experimentally introduced markers. In most cases, these marks are then read out by sequencing, discarding the spatial information of the cells. To address this vital gap in our toolkit, we developed a new synthetic lineage tracking system that allows us to image single-cell lineage history. This system, termed integrase-editable memory by engineered mutagenesis with optical in situ readout (intMEMOIR), uses serine integrases to stochastically and irreversibly edit a synthetic memory array, generating up to 59,049 different outcomes that can be unambiguously distinguished by fluorescence in situ hybridization (FISH). We evaluated the reconstruction accuracy of our system in mouse embryonic stem (mES) cells and disentangled the relative contribution of lineage and space to cell fate determination in Drosophila brain development, establishing the foundation for an expandable synthetic microscopy-readable system. In this thesis, Chapter 1 introduces the importance of cell lineage and spatial organization to cell fate determination, and includes a brief history of the existing technologies of the lineage tracking field. Chapter 2 describes our characterization and demonstration of the intMEMOIR system. Finally, Chapter 3 discusses design principles for robust, serine-integrase-based recording systems and suggests future directions for intMEMOIR
Biological Intelligence: from Behavior to Learning Theory
Get PDFKnowing how to learn, think, and act is not just a hallmark of intelligence, but a necessity of survival for many organisms. Behavior, the complete set of actions of species, allows us to glimpse into the minds of humans and animals, and by extension, intelligence itself. Biological intelligence is characterized by fast adaptation to changes and challenges, which is what allows species to survive in natural environments from starvation and predation. To study learning in a controlled setting, we can observe the behavior evoked through decision-making tasks that make it possible to quantify and analyze learning. By modeling the extracted behavioral features, we could start to understand the possible underlying mechanisms by proposing neural theory models, and look for those signals in the brain. Understanding the neural mechanisms of learning also strengthens the basis for building intelligent machines that are flexible and adaptive to the nonstationary world we live in. In this thesis, I present works in (1) automating behavioral setups and modeling suboptimal behavior in a traditional decision-making task, (2) using an ethological navigation task to characterize fast-sequence learning, and (3) how neural theory can explain some core behavioral phenomena in (2), and be used to solve a central problem in graph search.</p
Computational Methods for Simulating and Parameterizing Nucleic Acid Secondary Structure Thermodynamics and Kinetics
Get PDFNucleic acid secondary structure models offer a simplified but powerful lens through which to view, analyze, and design nucleic acid chemistry. Computational approaches based on such models are central to current research directions across molecular programming, synthetic biology, and the life sciences more broadly. Our framework combines three ingredients. First, we develop new recursions to include contributions from coaxial and dangle stacking in an efficient and principled way. Second, we formulate the concept of an evaluation algebra, which defines the mathematical form of each subproblem in the dynamic program. Whereas previous modeling efforts have relied on case-by-case handling of different thermodynamic quantities, we use evaluation algebras to elegantly and efficiently compute a variety of physical quantities using the same recursions. Third, we develop efficient operation orders for a variety of physical quantities of experimental interest. Combining our advances, we are able to achieve speedups of 20-120x and scalable calculations of complexes of up to 30,000 nucleotides. Our achievements promise to dramatically expand the scope and utility of computational analysis and design of nucleic acid thermodynamics.
While current dynamic programming algorithms achieve efficient computation of thermodynamic quantities for a given nucleic acid sequence, they do not provide kinetic information. Therefore, investigations of secondary structure kinetics rely on stochastic simulations of trajectories in secondary structure space. We improve upon these simulation methodologies to achieve lower computational complexities and large empirical speedups. We extend our algorithms to an ensemble which fully includes coaxial and dangle stacking states, expanding the scope of the kinetic analysis that is currently possible.
Current secondary structure models are parametrized using thermodynamic information gleaned from decades of melt experiments of RNA and DNA in specific experimental conditions. Only rough kinetic information is currently available from past experiments, and information on solvent and material dependence is lacking. We develop a fully computational approach based on Gaussian processes and molecular dynamics in order to provide a generic method for estimating thermodynamic and kinetic parameters, applicable conceptually to any nucleic acid material and experimental setting of interest. Our methodology offers an atomistic view of nucleic acid base pairing and faithfully reproduces most experimental data. It thus provides a powerful black-box approach for extensibly calculating the kinetic and thermodynamic parameters that secondary structure models require.</p
Radiation-Based Analytic Approaches to Investigate the Earth’s Atmosphere
Get PDFRadiation, propagating through Earth’s atmosphere, plays an important role in the Earth system. Solar radiation is the major source of energy, followed by thermal infrared radiation emitted by the Earth. The total radiative energy budget affects dynamic, thermodynamics, photochemical and biological processes. In addition, by measuring the reflected and emitted radiation at a distance (e.g., satellite or aircraft), we can detect and monitor the physical characteristics of a region which can help researchers get a better understanding of Earth’s atmosphere. Therefore, radiation-based analytic approaches are powerful tools in Earth Science. This thesis focuses on using radiation-based analytic tools to study the Earth’s atmosphere and to understand human impacts on the Earth system.
First, we develop novel machine learning methods for hyperspectral radiative transfer simulations. Hyperspectral technique is one of the most popular and powerful methods for atmospheric remote sensing and is widely used for temperature, gas, aerosol, and cloud retrievals. However, accurate forward radiative transfer simulations are computationally expensive since they require a larger number of monochromatic radiative transfer calculations. We, therefore explore the feasibility of machine learning techniques for fast hyperspectral radiative transfer simulations that perform calculations at a small fraction of hyperspectral wavelengths and extend them across the entire spectral range. The machine learning-based approach achieves better performance than the traditional principal component analysis (PCA) method.
Second, we evaluate modeled hyperspectral infrared spectra against satellite all-sky observations. The national weather centers obtain data from hyperspectral infrared sounders on a global scale. The cloudless scenario of this data is used to initialize weather forecasts, including temperature, water vapor, water cloud, and ice cloud profiles on a global grid. Although the data from these satellites are sensitive to the vertical distribution of ice and liquid water in the clouds, this information is not fully utilized. In this study, we evaluate how well the modeled spectra compare to AIRS observations using different cloud overlap models. We hope that this information can be used to verify clouds in the National Meteorological Center model and to initialize forecasts in the future.
In the last chapter, we use radiation-based analytic approaches to study human impacts on the Earth system. In the first study case, we show that the radiative forcing due to geospatially redistributed anthropogenic aerosols mainly determined the spatial variations of winter extreme weather in the Northern Hemisphere during 1970-2005, which is a unique transition period for global aerosol forcing. In the second case, we review satellite and ground-based observations and conduct state-of-art atmospheric model simulations during the COVID-19 lockdown period. The halted human activities during the COVID-19 pandemic in China provided a unique experiment to assess the efficiency of air-pollution mitigation.</p
Defect-Driven Reactivity of Layered Materials Examined through Atomic Layer Deposition
Get PDFLayered materials are a unique class of materials that are characterized by strong covalent bonds in single layers in the two-dimensional plane, but much weaker van der Waals interactions between layers. This unique bonding environment gives rise to remarkable chemical and physical properties. These materials can be easily exfoliated into single atomic layers, true two-dimensional materials, or few-layer stacks. As the number of layers changes, the physical and chemical properties can be modified. Similarly, as the relative position of one or more layers is changed, this change in the bonding environment has substantial impacts on the properties. Due to their unique bonding environment, layered materials are unusually inert to atomic layer deposition. Atomic layer deposition (ALD) is a surface-sensitive deposition technique that relies on a sequence of chemical reactions on the surface of a substrate to deposit the target material. Using this surface-sensitive deposition technique, the crystal structure and chemical bonding environment of the layered material can be interrogated.
Chapter 1 investigates the spontaneous formation of highly ordered triangular and linear pattern depositions on layered material substrates. These patterns form with two different layered material substrates and with two separate ALD reactions. The pattern depositions do not change with increasing deposition time or with different concentrations of reactants. These networks, while highly unusual to observe chemically, are discussed through a well-established dislocation theory, where defects in the crystal structure of layered materials can impact the surface reactivity.
Chapter 2 explores the stacking order characteristics of few-layer materials and discusses how changing the stacking order can change the crystal structure and the electronic properties of the materials. The chapter describes the field of few-layer materials and the process of modifying or transferring nanoflakes to new substrates. The chapter introduces a new transfer system that allows for patterned nanoflakes with stacking fault networks to be transferred to new substrates with micron-scale precision. The crystal structure of both the deposited triangular networks and the layered material beneath is discussed.
Chapter 3 describes a new approach to selectively targeting defect sites in monolayer graphene. A new, water-free atomic layer deposition chemistry is introduced to precisely react metal oxides with high-energy defect sites on graphene. The quality of the film is interrogated and the selectivity of the film is determined by measuring the thickness of the film deposited on the defect-rich regions. The results confirm that this ALD process creates a robust passivating film while keeping the unperturbed regions clean of any metal oxide.</p
Sparse Neural and Motor Networks Underlying Control in the Drosophila Flight System
Get PDFWe often look to the natural world for inspiration in design and engineering. The fruit fly, Drosophila melanogaster, with approximately 100,000 neurons in its central nervous system (CNS) versus the roughly 100 billion neurons of the human brain, is relatively uncompromising in the richness of behaviors it is capable of performing given its comparatively sparse nervous system. It exhibits exceptional aerial agility, despite the steep aerodynamic constraints of miniaturization thanks to unique physiological and biomechanical thoracic adaptations. However, the mechanisms governing its sparse and precise flight control have remained largely inaccessible due to technological and geometric limitations, leaving many long-standing questions in the field of insect flight control unexplored. Recent advances in the field of molecular biology have created a vast toolkit for both optical imaging and genetic manipulation of cellular function. This revolution of genetic advances allows us to visualize changes in muscle activity in situ as fluorescent signals, to record from fluorescently targeted cells via electrophysiology or 2-photon imaging, and to optogenetically activate or silence the activity of targeted cells. This thesis utilizes recent technological and molecular advances to probe three key aspects of fly flight control: 1) the dynamic interactions of flight steering muscles to produce flight maneuvers, 2) the source of timing information for the structuring of the the motor phase code, an extremely temporally precise wingbeat-synchronous aspect neural firing, and 3) the mechanisms by which slow, graded descending visual process recruit the flight muscles.
In the contents of the ensuing chapters I propose mechanisms for flight control pertaining to the wing muscles as well as their inputs. First, I describe the activities of each of the flight steering muscles in response to visual motion to generate movement in yaw, pitch, and roll (Chapter II). I then characterize the flexible individual dynamics and combinatorial timing of the system, and propose specific mechanisms by which interneurons rather than muscle physiology govern these adaptable firing patterns according to sensory inputs(Chapter II). Sensory inputs within this thesis take two forms: thoracic mechanosensory and timing information as well as descending visual input. I characterize mechanosensory and timing adaptations of an evolutionarily evolved hind wing, as well as the impact of haltere feedback to flight control (Chapter III). Lastly, I propose a mechanism by which descending visual commands produce graded outputs of the muscles.</p
Investigation of Electronic Fluctuations in Semiconductor Materials and Devices through First-Principles Simulations and Experiments in Transistor Amplifiers
Get PDFElectronic noise, or stochasticity in the current, voltage, and frequency of a carrier signal is caused by microscopic fluctuations in the occupation of quantum electronic states. In the context of scientific instrumentation, understanding the physical origin of these fluctuations is of paramount importance since the associated stochasticity ultimately limits the fidelity of information transmitted through electronically processed-signals. The unifying theme of the work presented in this thesis is the study of electronic fluctuations in semiconductor materials and devices. Our interest in this topic is twofold. First, while the Nyquist law dictates the equivalence of noise and transport properties for systems in thermal equilibrium, this relationship breaks down for systems driven out of equilibrium by external forcing. Simulating non-equilibrium electronic fluctuations can therefore provide new insights into the microscopic processes that control energy and momentum relaxation which would not be available from conventional studies of transport alone. Furthermore, because noise properties are sensitive to the microscopic details of the bandstructure and scattering, ab initio simulations of noise observables provide a more rigorous test of the accepted theory of charge transport and carrier scattering in materials. Second, cryogenic low noise amplifiers based on high electron mobility transistors (HEMTs) are widely used in electromagnetic detector chains in applications such as radio astronomy, deep space communications, and quantum computing. The design and optimization of HEMT devices have conventionally relied upon empirical circuit-level models of fluctuations in devices. As the noise performance of modern low-noise amplifiers has saturated to levels five to ten times above the standard quantum limit, these empirical models are unable to resolve the microscopic origin of the limiting excess noise. Identifying the microscopic mechanisms underpinning noise in modern amplifiers is therefore necessary to produce better devices for scientific instrumentation. In this work, we investigate electronic noise in semiconductor materials and devices with a combination of first-principles simulations and Schottky thermometry experiments in transistor amplifiers.
First, we present our work on the development of novel parameter-free simulations of non-equilibrium noise in semiconductor materials. While the ab initio theory of low-field electronic transport properties such as carrier mobility is well-established, an equivalent treatment of electronic fluctuations about a non-equilibrium steady state has remained less explored. Starting from the Boltzmann Transport Equation, we develop an ab initio method for hot electron noise in semiconductors. In contrast with the typical numerical methods used for electronic noise such as Monte Carlo techniques, no adjustable parameters are required in the present formalism with the electronic band structure and scattering rates calculated from first-principles. Our formalism enables a parameter-free approach to probe the microscopic transport processes that give rise to electronic noise in semiconductors. Next, we apply the developed method to compute the spectral noise power in two materials of technological interest, GaAs and Si. In our first study in GaAs, we show that despite the well-known dominance of optical phonon scattering, the spectral features in AC transport properties and noise originate from a surprising quasi-elasticity in the scattering of warm electrons with the lattice. In our second study, we apply the method to Si which possesses a more complicated multivalley conduction band. This study demonstrates that the widely-accepted one-phonon scattering approximation is insufficient to reproduce the warm electron tensor and that incorporating second-order mechanisms, such as two-phonon scattering, may be critical to obtain an accurate description of noise in such materials.
Finally, we discuss our work on developing deeper understandings of electronic noise in real devices with a focus on transistor amplifiers. While the first-principles work described above is appropriate for evaluating noise in ideal materials, in real semiconductor devices, charge carriers are influenced by mechanisms such as defect scattering, size effects, and reflections at interfaces. Owing to the complexity of these mechanisms, HEMT noise is typically treated with empirical models, where the physical noise sources are reduced to fitting parameters. Existing models of HEMT noise, such as the Pospieszalski model, are unable to resolve the mechanisms that set the noise floor of modern transistor amplifiers. In particular, the magnitude of the contribution of thermal noise from the gate at cryogenic temperatures remains unclear owing to a lack of experimental measurements of thermal resistance under these conditions. We report measurements of gate junction temperature and thermal resistance in a HEMT at cryogenic and room temperatures using a Schottky thermometry method. Based on our findings, we develop a phonon radiation model of heat transfer in the device and estimate that the thermal noise from the gate is several times larger than previously assumed. Our work suggests that self-heating results in a practical lower limit for the microwave noise figure of HEMTs at cryogenic temperatures.</p
New-to-Nature Selective C-H Alkylation Using Engineered Carbene Transferases
Get PDFSynthetic methods to selectively convert C–H bonds, a prevalent motif in organic molecules, into functionalities can significantly accelerate the syntheses and derivatization of molecules. In the past decade, many enzymatic catalysts have emerged as greener, more selective, and more versatile alternatives to small-molecule catalysts for selective C–H functionalization reactions. In nature, enzymes only catalyze a limited set of C–H functionalization reactions that are useful for chemical synthesis, as the overwhelming majority of known C–H functionalization enzymes in nature are hydroxylases. The diversity of the enzymatic reaction scope needs to be substantially expanded to make them broadly useful to synthetic chemists. This thesis will describe new enzymes, which we repurposed from one of the most prevalent C–H hydroxylases in nature, cytochromes P450, which are now able to catalyze new-to-nature C–H alkylation reactions via selective carbene transfer. Given the central role of C–C bond forming reactions in building and elaborating the carbon skeleton of organic molecules, these transformations are of high interest in many fields of research, such as medicinal chemistry and material chemistry. In Chapter 1, I review a recent surge in newly identified enzymes, repurposed enzymes, and artificial metalloenzymes which can catalyze selective C–H functionalization reactions. Chapter 2 details the development of a panel of enantiodivergent α-amino C(sp³)−H fluoroalkylases. Using directed evolution, the carbene transferases can install fluoroalkyl groups onto these C–H bonds with high activity (4,070 total turnovers, TTN) and selectivity (>99% ee). Notably, complementary regioselectivity can be achieved using an alternative enzyme, P411-PFA-(S). In Chapter 3, I report the first carbene transferase, P411-ACHF, which can transfer an α-cyanocarbene to arene C–H bonds of N-substituted benzenes. Chemodivergent C(sp²)–H and C(sp³)–H functionalization can be achieved using P411-ACHF and P411-PFA. Additionally, structural studies revealed an unprecedented backbone carbonyl flip within the long I-helix of P411-PFA, which may suggest how these enzymes have evolved to bind and activate diazo compounds for carbene transfer reactions. In Chapter 4, I discuss the efforts I took toward stabilizing an interesting but unstable P450, CYP3A4. This enzyme exhibits large active site volume and high substrate promiscuity and therefore can be a great candidate to develop late-stage carbene and nitrene transferases. I adopted consensus sequence mutagenesis and predicted five mutations which have the potential to have the strongest beneficial effects on improving CYP3A4's thermostability. In summary, this thesis work addresses the urgent need for expansion of the current enzymatic C–H functionalization reaction scope and the development of more sustainable and selective C–H functionalization catalysts which can be synthetically useful.</p
Nickel-Catalyzed Electroreductive Cross-Coupling Reactions of Anhydrides and Alkyl Halides
No full textThe formation of new carbon–carbon bonds is one of the most important transformations in organic chemistry due to its ability to build the backbone of organic molecules. Nickel-catalyzed reductive cross-coupling reactions have recently emerged as an efficient and powerful strategy for the creation of new carbon–carbon bonds. Furthermore, electrochemistry can be harnessed to overcome some of the challenges encountered in many of the reductive cross-coupling reactions in the literature. Herein, we discuss the development of a new electroreductive nickel-catalyzed cross-coupling of anhydrides with unactivated alkyl bromides in collaboration with Amgen to produce large amounts of substituted cyclobutane products
Autonomous Mission-Driven Robots in Extreme Environments
Get PDFRobotic autonomy systems that can negotiate harsh environments under time and communication constraints are critical to accomplishing many real-world missions. Such systems require an integrated software-hardware solution capable of robustly reasoning about a time-limited mission across a complex environment and negotiating extreme physical conditions during mission execution. To this end, I will discus the development of two field-tested systems designed for operation in GPS-denied areas: (i) a coverage planning framework that enables efficient exploration of large, unknown environments, and (ii) a ballistically-launched aircraft that converts to an autonomous, free-flying multirotor in order to provide rapid aerial surveillance.
The first system addresses the time-limited exploration problem by providing a planning strategy that seeks to maximize the area covered by a robot’s sensor footprint along a planned trajectory. In order to find solutions over large spatial extents (>1 km) and long temporal horizons (>1 hour), this coverage problem is decomposed into tractable subproblems by introducing spatial and temporal abstractions. Spatially, the robot-world belief is approximated by a task-dependent structure, enriched with environment map estimates. Temporally, the belief is approximated by the aggregation of multiple structures, each spanning a different spatial range. Cascaded uncertainty-aware solvers return a coverage plan over the stratified belief in real time.
Coverage policies are constructed in a receding horizon fashion to ensure motion smoothness and resiliency to real-world stochasticity in perception and control. This coverage planning framework was extensively tested on physical robots in various real-world environments (caves, mines, subway systems, etc.) and served as the exploration strategy for a competing entry in the DARPA Subterranean Challenge.
The second system addresses rapid multirotor deployment for aerial data collection during emergencies. While multirotors are advantageous over fixed-winged systems due to their high maneuverability, their rotating blades are hazardous and require stable, uncluttered takeoff sites. To overcome this issue, a ballistically-launched, autonomously-stabilizing multirotor (SQUID -- Streamlined Quick Unfolding Investigation Drone) was designed, fabricated, and tested. SQUID follows a deterministic trajectory, transitioning from a folded launch configuration to an autonomous, fully-controllable hexacopter. The entire process from launch to position stabilization requires no user- or GPS-input and demonstrates the viability of using ballistically-launched multirotors to achieve safe and rapid deployment from moving vehicles.</p