12023 research outputs found
Sort by
Enhanced Algorithms for Analysis and Design of Nucleic Acid Reaction Pathways
Get PDFNucleic acids provide a powerful platform for programming at the molecular level. This is possible because the free energy of nucleic acid structures is dominated by the local interactions of base pairing and base pair stacking. The nearest neighbor secondary structure model implied by these energetics has enabled development of a set of algorithms for calculating thermodynamic quantities of nucleic acid sequences. Molecular programmers and synthetic biologists continue to extend their reach to larger, more complicated nucleic acid complexes, reaction pathways, and systems. This necessitates a focus on new algorithm development and efficient implementations to enable analysis and design of such systems.
Concerning analysis of nucleic acids, we collect seemingly diverse algorithms under a unified three-component dynamic programming framework consisting of: 1) recursions that specify the dependencies between subproblems and incorporate the details of the structural ensemble and the free energy model, 2) evaluation algebras that define the mathematical form of each subproblem, 3) operation orders that specify the computational trajectory through the dependency graph of subproblems. Changes to the set of recursions allows operation over the complex ensemble including coaxial and dangle stacking states, affecting all thermodynamic quantities. An updated operation order for structure sampling allows simultaneous generation of a set of structures sampled from the Boltzmann distribution in time that scales empirically sublinearly in the number of samples and leads to an order of magnitude or more speedup over repeated single-structure sampling.
For the problem of sequence design for reaction pathway engineering, we introduce an optimization algorithm to minimize the multitstate test tube ensemble defect, which simultaneously designs for reactant, intermediate, and product states along the reaction pathway (positive design) and against crosstalk interactions (negative design). Each of these on-pathway or crosstalk states is represented as a target test tube ensemble containing arbitrary numbers of on-target complexes, each with a target secondary structure and target concentration, and arbitrary numbers of off-target complexes, each with vanishing target concentration. Our test tube specification formalism enables conversion of a reaction pathway specification into a set of target test tubes. Sequences are designed subject to a set of hard constraints allowing specification of properties such as sequence composition, sequence complementarity, prevention of unwanted sequence patterns, and inclusion of biological sequences. We then extend this algorithm with soft constraints, enhancing flexibility through new constraint types and reducing design cost by up to two orders of magnitude in the most highly constrained cases. These soft constraints enable multiobjective design of the multitstate test tube ensemble defect simultaneously with heuristics for avoiding kinetic traps and equalizing reaction rates to further aid reaction pathway engineering.</p
Concise Total Syntheses of ∆¹²-Prostaglandin J Natural Products Using Stereoretentive Metathesis
Get PDF∆¹²-Prostaglandin J family is a type of secondary metabolite isolated in cell culture, and is recently discovered to have potent anticancer activity. Concise syntheses of four ∆¹²-prostaglandin J natural products (7–8 steps in the longest linear sequences) are developed, enabled by convergent stereoretentive cross-metathesis by Ru-based metathesis catalyst. Exceptional control of alkene geometry was achieved through stereoretention.</p
Nanomechanical Properties of Electrodeposited Li and Fabrication of 3D Architected Cathodes for Li-Based Batteries
Get PDFAdvancements in the active materials of Li-based batteries provide a promising route to significantly improve electrochemical performance. Li metal has a 10x increase in gravimetric capacity compared to conventional graphite anodes and can be utilized with a solid electrolyte. However, current solid-state Li metal anode batteries cannot reliably cycle large amounts of Li due to chemical and mechanical degradation at the solid electrolyte / Li interface. One key factor in the failure of solid electrolytes is the dearth of mechanical data on Li at the relevant length scales and microstructures to solid-state batteries. The initial stages of Li formation at the solid electrolyte / Li interface also require further exploration to help improve the performance of solid-state Li batteries.
In the first part of the thesis, we will discuss the methods used to investigate Li electrodeposited in-situ in a scanning electron microscope (SEM) chamber from a thin film solid-state battery. We probed the formation of this Li and found preferential growth at the domain boundaries of the surface of the cell, corroborated by electrochemical simulations. Cryogenic electron microscopy was determined to be the optimal method for examining the microstructure of Li and was utilized to reveal the single crystalline microstructure of Li pillars. Uniaxial compression experiments were performed on single crystalline Li pillars that grew from these batteries. We found that Li pillars with diameters of 360-759 nm first deformed elastically, then yielded and flowed plastically, with an average yield stress of 16.0 ± 6.82 MPa, 24x stronger than bulk polycrystalline Li. The mechanical results are discussed in the framework of dislocation starvation and nucleation, in addition to thermally activated deformation processes.
Next generation battery systems may also utilize 3D electrodes to allow for both high energy (large mass loading) and power densities (small diffusion lengths). The last section of the thesis investigates the fabrication of 3D architected LiCoO2 structures and their performance as Li-ion battery cathodes. Using a novel hydrogel photoresin with relevant salt contents, the structures were fabricated using digital light processing and calcination. The electrochemical performance of the architected cathodes was examined and the electrodes exhibited a relatively high areal capacity up to ∼8 mAh/cm2 and a capacity retention of 82% after 100 cycles.</p
A Quantitative and High-Throughput Approach to Gene Regulation in Escherichia coli
Get PDFMeasurements in biology have reached a level of precision that demands quantitative modeling. This is particularly true in the field of gene regulation, where concepts from physics such as thermodynamics have allowed for accurate models to be made.
Many issues remain. DNA sequencing is routine enough to sequence new genomes in days and cheap enough to use deep sequencing to perform precision measurements, but our ability to interpret the wealth of genomic data is lagging behind, especially in the realm of gene regulation. The primary reason is that we lack any information what so ever as to the basic regulatory details of approximately 65 percent of operons even in E. coli, the best understood organism in biology. As a result we cannot use our hard won modeling efforts to understand any of these operons.
This work takes steps to address these issues. First we use 30 LacI mutants as a test case to prove that we can make quantitatively accurate models of gene expression and sequence-dependent binding energies of transcription factors and RNA polymerase.
Next we note that much of the quantitative insight available on transcriptional regulation relies on work on only a few model regulatory systems such as LacI as was considered above. We develop an approach, through a combination of massively parallel reporter assays, mass spectrometry, and information-theoretic modeling that can be used to dissect bacterial promoters in a systematic and scalable way. We demonstrate that we can uncover a qualitative list of transcription factor binding sites as well as their associated quantitative details from both well-studied and previously uncharacterized promoters in E. coli.
Finally we extend the above method to over 100 E. coli promoters using over 12 growth conditions. We show the method recapitulates known regulatory information. Then, we examine regulatory architectures for more than 80 promoters which previously had no known regulation. In many cases, we identify which transcription factors mediate their regulation. The method introduced clears a path for fully characterizing the regulatory genome of E. coli and advances towards the goal of using this method on a wide variety of other organisms including other prokaryotes and eukaryotes such as Drosophila melanogaster.</p
TIME: A Millimeter-Wavelength Grating Spectrometer Array for [CII] / CO Intensity Mapping
Get PDFIn this thesis I review the design, fabrication, and initial engineering deployment of the TIME (Tomographic Ionized-carbon Mapping Experiment) instrument. TIME seeks to make a first detection of the clustering amplitude of the power spectrum of redshifted [CII] emission from the Epoch of Reionization (z = 5-9). [CII], the 157.7 µm fine-structure line of singly ionized carbon, traces star formation on large scales, providing a new method for constraining the contribution of star formation to the Reionization process. [CII] intensity mapping complements traditional galaxy surveys by using spatially-broad beams to integrate signal from the many faint sources thought to be responsible for the bulk of the integrated emission from galaxies. TIME covers the 200-300 GHz atmospheric window, which also enables the study of lower-redshift CO emission (z = 0.5-2), a tracer of molecular gas in the period following the peak of cosmic star formation. The full TIME instrument consists of 32 single-polarization grating spectrometers with a resolution R ~ 100. Each spectrometer consists of an input feedhorn coupled to parallel plate waveguide with a curved diffraction grating, which focuses the diffracted light onto an output arc populated by 60 transition-edge sensor (TES) bolometers at 250 mK. The 1920 total detectors couple to the output of the parallel plate waveguide with a direct-absorbing micro-mesh and are organized into buttable arrays covering 4 spatial by either 12 (HF) or 8 (LF) spectral pixels. A partial TIME instrument was field tested in early 2019 on the ARO APA 12m dish at Kitt Peak. We intend to return to Kitt Peak in late 2020 to begin initial science observations.</p
Positive Definite Matrices: Compression, Decomposition, Eigensolver, and Concentration
Get PDFFor many decades, the study of positive-definite (PD) matrices has been one of the most popular subjects among a wide range of scientific researches. A huge mass of successful models on PD matrices has been proposed and developed in the fields of mathematics, physics, biology, etc., leading to a celebrated richness of theories and algorithms. In this thesis, we draw our attention to a general class of PD matrices that can be decomposed as the sum of a sequence of positive-semidefinite matrices. For this class of PD matrices, we will develop theories and algorithms on operator compression, multilevel decomposition, eigenpair computation, and spectrum concentration. We divide these contents into three main parts.
In the first part, we propose an adaptive fast solver for the preceding class of PD matrices which includes the well-known graph Laplacians. We achieve this by establishing an adaptive operator compression scheme and a multiresolution matrix factorization algorithm which have nearly optimal performance on both complexity and well-posedness. To develop our methods, we introduce a novel notion of energy decomposition for PD matrices and two important local measurement quantities, which provide theoretical guarantee and computational guidance for the construction of an appropriate partition and a nested adaptive basis.
In the second part, we propose a new iterative method to hierarchically compute a relatively large number of leftmost eigenpairs of a sparse PD matrix under the multiresolution matrix compression framework. We exploit the well-conditioned property of every decomposition components by integrating the multiresolution framework into the Implicitly Restarted Lanczos method. We achieve this combination by proposing an extension-refinement iterative scheme, in which the intrinsic idea is to decompose the target spectrum into several segments such that the corresponding eigenproblem in each segment is well-conditioned.
In the third part, we derive concentration inequalities on partial sums of eigenvalues of random PD matrices by introducing the notion of k-trace. For this purpose, we establish a generalized Lieb's concavity theorem, which extends the original Lieb's concavity theorem from the normal trace to k-traces. Our argument employs a variety of matrix techniques and concepts, including exterior algebra, mixed discriminant, and operator interpolation.</p
Development of Enantioselective Transition-Metal Catalyzed Allylic Alkylation Methodologies
Get PDFResearch in the Stoltz group is directed, generally, at the development of synthetic methods for the preparation of stereochemically rich molecules, and the total synthesis of complex natural products. One major theme of our group’s methods development is transition-metal catalyzed allylic alkylation, of which we have reported Pd, Ir, Cu, and Ni catalyzed strategies. Described in this thesis are projects related to these interests, primarily focused on new approaches toward acyclic stereocenters via palladium catalysis; however also include an iridium-catalyzed formal γ-alkylation of malonates and β-ketoesters, as well as an Overman rearrangement strategy for synthesizing α-amino ketones. A majority of asymmetric enolate functionalization methods, developed by our group and others, pertain to cyclic systems in which only one enolate geometry isomer is possible due to the constrained ring. In acyclic systems, however, this issue of enolate geometry becomes a major challenge that must be addressed. When one seeks to prepare a tetrasubstituted acyclic enolate, which would lead to a fully-substituted stereocenter following functionalization, one must contend with the issue of non-selective formation of a mixture of enolates which generally leads toward less selective transformations. Strategies toward overcoming this issue, as well as new insights gained regarding the palladium-catalyzed alkylation of acyclic enolates, are described in the subsequent chapters.</p
Modeling and Development of Superconducting Nanowire Single-Photon Detectors
Get PDFSuperconducting nanowire single-photon detectors (SNSPDs) have demonstrated remarkable efficiency, timing resolution, and intrinsic dark count rate properties, but the SNSPD community currently lacks a comprehensive model of the single-photon detection process. In this work, we conduct a detailed examination of the current detection mechanism models and compare their predictions to new experimental measurements of the intrinsic timing properties and polarization dependence of specialized NbN test devices. First, we consider the energy downconversion cascade using the kinetic equations to describe the non-equilibrium electron and phonon systems immediately following photon absorption. These calculations provide estimates for the energy loss and fluctuations during this process, and provide qualitative information about the way energy is partitioned between the electron and phonon systems. To study the suppression of superconductivity following downconversion, we apply the most advanced existing model, that of Vodolazov (2017), but find it inadequate to quantitatively describe the timing properties of these detectors. By extending the model to use the generalized time-dependent Ginzburg-Landau equations, we achieve better quantitative agreement with experiment. However, the generalized model still provides only a qualitative picture of the detection process.
We also conduct an experimental examination of the heat transfer process in WSi nanowires by examining the nanowire reset dynamics, steady-state dissipation, and crosstalk between elements of an array. The results are compared to existing electrothermal models, but these models fail to adequately describe the dynamics of the system. A generalized form of the electrothermal model provides better fitting to experiment, but incorporation of non-equilibrium effects is likely needed to provide a fully quantitative description of the system. These results are directly connected to some of the thermal challenges of SNSPD array development. Informed by the crosstalk results, we demonstrate a new multiplexing technique based on thermal coupling between two active nanowire layers, known as the thermal row-column. This method promises to enable kilopixel to megapixel scale imaging arrays for low photon-flux applications. Finally, we discuss the design and characterization of the ground detector for the Deep Space Optical Communication (DSOC) demonstration mission.</p
C(sp³)–H Activation via Dehydrogenation of Cyclic and Heterocyclic Alkanes by Single-Site Iridium Pincer Ligated Complexes
Get PDFThe direct dehydroaromatization of C(sp³)–H alkanes may seem conceptually simple but in fact is a challenging transformation. Industrially practiced methods utilize energy intensive processes operating at high pressures and temperatures due to the requirement of such conditions to overcome the endergonic and unreactive nature of alkanes. Chapter 1 briefly discusses early and recent achievements in the field of alkanes dehydrogenation by Ir pincer ligated complexes. While there has been great advancement in the dehydrogenation transformation recently, the direct dehydroaromatization of heterocyclic substrates generating functionalized aromatics is significantly underdeveloped. In Chapter 2, we successfully extended the applicability of Ir catalyzed dehydrogenation systems using pincer ligated complexes on a diverse collection of heterocyclic alkanes with functionalities known to be strongly coordinating and poorly compatible with (PCP)–Ir type catalysts. Carbo- and heteroarenes containing oxygen and nitrogen can be synthesized in moderate to excellent yields up to 99%, and the reaction tolerates functional groups such as bromides and fluorides. In Chapter 3, we demonstrate the efficient disproportionation of cycloalkenes to the corresponding arenes and cycloalkanes with up to 100% conversion, which has been a long-standing challenge in the field of pincer-ligated Ir-catalyzed dehydrogenation studies. For example, 1-cyclohexene was disproportionated to benzene and cyclohexane and 1-4-vinyl-1-cyclohexene was disproportionated to ethylbenzene and ethylcyclohexane. We also demonstrate that a key mechanistic feature of our system is a lack of catalyst inhibition by arenes. In addition, our method is advantageous to previous reports as no sacrificial olefin is used, thereby circumventing the requirement for exogeneous hydrogen acceptors. Our studies presented in Chapter 2 and Chapter 3 provides a novel and a complementary pathway to access important aromatic building blocks and may help create alternative routes to complex molecules via late stage dehydrogenation without the need of stoichiometric oxidants
New Frameworks for Structured Policy Learning
Get PDFSequential decision making applications are playing an increasingly important role in everyday life. Research interest in machine learning approaches to sequential decision making has surged thanks to recent empirical successes of reinforcement learning and imitation learning techniques, partly fueled by recent advances in deep learning-based function approximation. However in many real-world sequential decision making applications, relying purely on black box policy learning is often insufficient, due to practical requirements of data efficiency, interpretability, safety guarantees, etc. These challenges collectively make it difficult for many existing policy learning methods to find success in realistic applications.
In this dissertation, we present recent advances in structured policy learning, which are new machine learning frameworks that integrate policy learning with principled notions of domain knowledge, which spans value-based, policy-based, and model-based structures. Our framework takes flexible reduction-style approaches that can integrate structure with reinforcement learning, imitation learning and robust control techniques. In addition to methodological advances, we demonstrate several successful applications of the new policy learning frameworks.</p