This dissertation considers the problem of learning the underlying statistical structure of complex data sets for fitting a generative model, and for both supervised and unsupervised data-driven decision making purposes. Using properties of exponential family distributions, a new unified theoretical model called Generalized Linear Statistics is established. The complexity of data is generally a consequence of the existence of a large number of components and the fact that the components are often of mixed data types (i.e., some components might be continuous, with different underlying distributions, while other components might be discrete, such as categorical, count or Boolean). Such complex data sets are typical in drug discovery, health care, or fraud detection. The proposed statistical modeling approach is a generalization and amalgamation of techniques from classical linear statistics placed into a unified framework referred to as Generalized Linear Statistics (GLS). This framework includes techniques drawn from latent variable analysis as well as from the theory of Generalized Linear Models (GLMs), and is based on the use of exponential family distributions to model the various mixed types (continuous and discrete) of complex data sets. The methodology exploits the connection between data space and parameter space present in exponential family distributions and solves a nonlinear problem by using classical linear statistical tools applied to data that have been mapped into parameter space. One key aspect of the GLS framework is that often the natural parameter of the exponential family distributions is assumed to be constrained to a lower dimensional latent variable subspace, modeling the belief that the intrinsic dimensionality of the data is smaller than the dimensionality of the observation space. The framework is equivalent to a computationally tractable, mixed data-type hierarchical Bayes graphical model assumption with latent variables constrained to a low- dimensional parameter subspace. We demonstrate that exponential family Principal Component Analysis, Semi- Parametric exponential family Principal Component Analysis, and Bregman soft clustering are not separate unrelated algorithms, but different manifestations of model assumptions and parameter choices taken within this common GLS framework. Because of this insight, these algorithms are readily extended to deal with the important mixed data -type case. This framework has the critical advantage of allowing one to transfer high-dimensional mixed-type data components to low-dimensional common-type latent variables, which are then, in turn, used to perform regression or classification in a much simpler manner using well-known continuous-parameter classical linear techniques. Classification results on synthetic data and data sets from the University of California, Irvine machine learning repository are presented

This music score was submitted for the Kaleidoscope 2020 Call for Scores, an open access collaboration with the UCLA Music Library.

This music score was submitted for the Kaleidoscope 2020 Call for Scores, an open access collaboration with the UCLA Music Library.

This music score was submitted for the Kaleidoscope 2020 Call for Scores, an open access collaboration with the UCLA Music Library.

Since the discovery of Mimivirus, viruses with large genomes encoding components of the translation machinery and other cellular processes have been described as belonging to the nucleocytoplasmic large DNA viruses. Recently, genome-resolved metagenomics led to the discovery of more than 40 viruses that have been grouped together in a proposed viral subfamily named Klosneuvirinae. Members of this group had genomes of up to 2.4Mb in size and featured an expanded array of translation system genes. Yet, despite the large diversity of the Klosneuvirinae in metagenomic data, there are currently only two isolates available. Here, we report the isolation of a novel giant virus known as Fadolivirus from an Algerian sewage site and provide morphological data throughout its replication cycle in amoeba and a detailed genomic characterization. The Fadolivirus genome, which is more than 1.5Mb in size, encodes 1,452 predicted proteins and phylogenetic analyses place this viral isolate as a near relative of the metagenome assembled Klosneuvirus and Indivirus. The genome encodes for 66 tRNAs, 23 aminoacyl-tRNA synthetases and a wide range of transcription factors, surpassing Klosneuvirus and other giant viruses. The Fadolivirus genome also encodes putative vacuolar-type proton pumps with the domains D and A, potentially constituting a virus-derived system for energy generation. The successful isolation of Fadolivirus will enable future hypothesis-driven experimental studies providing deeper insights into the biology of the Klosneuvirinae.

This paper presents the microfabrication and performance of a three-dimensional electrospray ionization (ESI) emitter tip made from glass, which achieves stable current signals important for chemical analysis. Our novel microfabrication process and custom-built signal conditioning hardware provides the advantage of providing accurate features and steady signals. The fabrication process relies on standard microfabrication techniques (i.e., deposition, photolithography, and wet etching). This fabrication method involves the novel application of two layers of positive and negative photoresists in addition to Parafilm^® wax tape. Open edge and tiered depth details were successfully created from a multilayer planar mask. This is a benefit for integrated miniaturized and microfluidic systems that often require micro features for their functionality but relatively large millimeter size features for their physical periphery. We demonstrate the fundamental performance of electrospray with this microfluidic chip. The emitter tip was fixed on a linear axis stage with high resolution (10 μm) to finely control the tip distance from a metal counter electrode plate. A custom printed circuit board system was built to safely control four voltages applied to the microchip ports from a single high voltage power supply. To readily form the electrospray, non-aqueous solvents were used for their low viscosity and a constant voltage of +2.7 kV was applied to the sheath electrospray microchannel. The liquid being sprayed was 80/20 (v/v) methanol/acetonitrile with 0.1% acetic acid in the sheath microchannel and with ammonium acetate (10-40 mM) in its remaining microchannels. The electrospray signal was measured in response to varying the distance (1.4 to 1.6 mm) between the electrospray emitter tip and a metal counter electrode plate in addition to the varying concentration of the background electrolyte, ammonium acetate. Stable and repeatable electrospray signal showed linear relationships with emitter tip distance and concentration (r²≥0.95).

Fungi of the phylum Basidiomycota (basidiomycetes), make up some 37percent of the described fungi, and are important in forestry, agriculture, medicine, and bioenergy. This diverse phylum includes the mushrooms, wood rots, symbionts, and plant and animal pathogens. To better understand the diversity of phenotypes in basidiomycetes, we performed a comparative analysis of 35 basidiomycete fungi spanning the diversity of the phylum. Phylogenetic patterns of lignocellulose degrading genes suggest a continuum between the white rot and brown rot modes of wood decay. Patterns of protein kinases, secondary metabolic enzymes, and secreted proteins shed additional light on the broad array of phenotypes found in the basidiomycetes. We suggest that the lignocellulose gene content of an organism can be used to predict its nutritional mode, and predict Dacryopinax sp. as a brown rot; Botryobasidium botryosum and Jaapia argillacea as white rots.

We have developed a simple-to-manufacture microfabricated gas preconcentrator for MEMS-based chemical sensing applications. Cavities and microfluidic channels were created using a wet etch process with hydrofluoric acid, portions of which can be performed outside of a cleanroom, instead of the more common deep reactive ion etch process. The integrated heater and resistance temperature detectors (RTDs) were created with a photolithography-free technique enabled by laser etching. With only 28 V DC (0.1 A), a maximum heating rate of 17.6 °C/s was observed. Adsorption and desorption flow parameters were optimized to be 90 SCCM and 25 SCCM, respectively, for a multicomponent gas mixture. Under testing conditions using Tenax TA sorbent, the device was capable of measuring analytes down to 22 ppb with only a 2 min sample loading time using a gas chromatograph with a flame ionization detector. Two separate devices were compared by measuring the same chemical mixture; both devices yielded similar peak areas and widths (fwhm: 0.032-0.033 min), suggesting reproducibility between devices.

Fungi of the phylum Basidiomycota (basidiomycetes), make up some 37percent of the described fungi, and are important in forestry, agriculture, medicine, and bioenergy. This diverse phylum includes the mushrooms, wood rots, symbionts, and plant and animal pathogens. To better understand the diversity of phenotypes in basidiomycetes, we performed a comparative analysis of 35 basidiomycete fungi spanning the diversity of the phylum. Phylogenetic patterns of lignocellulose degrading genes suggest a continuum rather than a sharp dichotomy between the white rot and brown rot modes of wood decay. Patterns of secondary metabolic enzymes give additional insight into the broad array of phenotypes found in the basidiomycetes. We suggest that the profile of an organism in lignocellulose-targeting genes can be used to predict its nutritional mode, and predict Dacryopinax sp. as a brown rot; Botryobasidium botryosum and Jaapia argillacea as white rots.