Small RNAs are a class of non-coding RNAs that mediate gene silencing at both transcriptional and post-transcriptional levels in eukaryotes. In plants, microRNAs (miRNAs) and small interfering RNAs (siRNAs) are two core types of small RNAs, which play essential roles in diverse biological processes. Both the metabolism and activities of small RNAs contribute to the regulation of their targets. Even though many on small RNA pathway have been reported, the detailed mechanism on its metabolism is still lacking. To address this issue, my thesis research has employed multidisciplinary approaches and techniques. The achieved results and conclusions can be closed as the following. First, it was found in this research that an Argonaute protein AGO10 represses the accumulation of miR165/6 without affecting its biogenesis. AGO10 overexpression results in a reduction in full-length miR165/6 species accompanied with an elevated 3' truncation of miR165/6, which is rescued by losses of two exonucleases SDN1 and SDN2. An enzymatic assay between SDN1 and AGO10 immunoprecipitates in vitro was carried out, which confirms that AGO10 represses the accumulation of miR165/6 through promoting its degradation via SDN1. A similar assay between SDN1 and AGO1 immunoprecipitates was also performed, allowing the proposal of a model on miRNA degradation. This is the work reported in the Chapter Two of this dissertation. Second, in this research, through small RNA sequencing and comparison to wild type, many siRNAs were identified to be differentially expressed in the rcd1 mutant (int51). A whole-genome bisulfate-sequencing reveals that DNA methylation in all sequence contexts is greatly reduced on euchromatic arms in int51, while is not changed at heterochromatic regions. It was further found that the global level of H3K9me2 is reduced in int51. The changes observed in int51 resemble those of mutants defective in DNA or histone methylation, providing a novel insightful hint to dissect the RCD1-mediated regulatory network further. My research towards elucidating this network is reported in Chapter Three. Taken together, my dissertation research has established new understanding on the mechanism of plant miRNA degradation and has also identified a novel candidate in plant siRNA metabolism.

The first phase selection and the phase formation sequence between metal and silicon (Si) couples are indispensably significant to microelectronics. With increasing scaling of device dimension to nano regime, established thermodynamic models in bulk and thin film fail to apply in one dimensional (1-D) nanostructures. Herein, we use a kinetic competition model to explain the phase formation sequence of 1-D nickel (Ni) silicides: multiple Ni silicides coexist at the initial stage and then the fastest one wins out as the first phase in a following growth competition. With kinetic parameters extracted from in-situ transmission electron microscope (TEM) observations, we quantitatively explain the unique size dependant first phase formation and the phase formation sequence changes in 1-D structures. We can further control the first phase by selectively enhancing or suppressing the growth rate of silicides through template structure modifications. Growth rate diffusion limited phases can be greatly enhanced in a porous Si nanowire (NW) template due to short diffusion paths. On the other hand, a thick aluminum oxide (Al₂O₃) shell around the NW is applied to impede the growth of large volume diffusion limited phases including Ni₃₁Si₁₂, theta-Ni₂Si and delta-Ni₂Si. Moreover, a thin platinum (Pt) interlayer between Si and Ni is used to suppress the nucleation of NiSi₂. Together, with the thick shell and Pt interlayer, we can suppress all competing silicides and render slow growing NiSi to form as the first phase. The resistivity of Pt doped NiSi (denoted as Ni(Pt)Si) NW are found compatible to pure NiSi from a two terminal and four terminal measurement. Controlled formation of Ni₃₁Si₁₂, theta-Ni₂Si, delta-Ni₂Si, NiSi or NiSi₂ as the first phase has also been achieved. To examine the kinetic competition model, 1-D cobalt (Co) and palladium (Pd) silicide formations are also studied and analyzed kinetically. A thick shell is found effective to suppress the Pd silicide NW broken at the interface.

Many algorithms in Machine Learning have demonstrated powerful empirical performance in terms of prediction accuracy and ability to extract knowledge from data. Inspired by the empirical success, researchers study the behavior of those algorithms under theoretical models. Two challenges must be addressed when one {tries} to understand those algorithms from a theoretical perspective. First, a sound theoretical model should be considered. A good model should reflect key properties of the real data. If the models do not capture important aspects of the real data, the observations and/or conclusions made on those theoretical models {may not be relevant} to practice. Thus, theoretical models that reflect certain properties of real data are required so that the insights obtained from those models are convincing to practitioners. Second, many of these algorithms are hard to analyze via traditional techniques. Therefore, novel techniques are required to analyze those algorithms under new theoretical models. In this thesis, we analyze two problems under novel real-data inspired theoretical models: $\ell_1$-minimization for dictionary learning and seeking important features and feature interactions from {Random Forest} (RF). Both analyses give unique insights into the problem by studying a novel data generative model. For dictionary learning, we propose two novel theoretical models: exact sparse model and Bernoulli-type models. Unlike most previous {analyses} that assume the data is {generated from} Gaussian distributions with sparsity constraints, these new models can capture non-Gaussian data distributions and allow us to analyze the algorithms under novel data properties such as non-negativity and heavy-tail. We show that $\ell_1$-minimization model in Dictionary Learning does not satisfy the classic global identifiability condition under the new model. However, the reference dictionary still enjoys some global property across all the feasible dictionaries. Our theoretical analysis leads to a novel algorithm called Dictionary Learning Block Coordinate Descent (DL-BCD). For RF, we start off with analyzing the feature importance bias for noisy features when using Mean Decrease Impurity (MDI). Then, we study the feature interaction recovery problem and analyzed the data-inspired Local-Spiky Sparse (LSS) model without Lipschitz assumptions that are often present in the previous literature. We show that the depth-weighted-prevalence of a true feature interaction in the decision paths of trees does not depend on the model coefficients but only on the size of the interaction. The theoretical analysis leads to a novel feature ranking method called LSSrank. We examine the performance of LSSrank on simulated data and it has high probability to rank true interactions at the top under the LSS model.

Classical transportation systems have been suffering from inefficiency in utilizing road capacity and incapability of guaranteeing safety due to inconsistent and unpredictable human based manipulations. This long standing issue is expected to be resolved through the introduction of vehicular autonomy, along with a properly configured traffic management system and an underlying communication network that supports the dissemination of safety and control messages. However, mechanisms to be used in the regulation of autonomous vehicles are yet to be developed.

For the first part of the dissertation, we propose an efficient data dissemination protocol, identified as Vehicular Backbone Network (VBN), which support high throughput and low delay message delivery across a Vehicular Ad Hoc Network (VANET). Vehicles are elected as relay nodes at properly selected nominal positions to optimize system throughput rate with guaranteed outage probability. The proposed VBN mechanism is analyzed under different Medium Access Control (MAC) protocols and various stochastic traffic flow conditions.

In the second part of the dissertation, we go beyond the statistical vehicular traffic model

and present our study of a platoon-based autonomous vehicular traffic regulation system. The traffic management system, serving to configure vehicles into platoons and regulating the on-ramp access of vehicular flows into the highway, is integrated with a design of a VANET system. The latter provides for the communications of essential message flows among highway vehicles. The fundamental tradeoffs that need to be made in designing the autonomous systems are modeled and studied. We propose an optimization framework to be used in designing the traffic control system in a manner that meet throughput and latency performance metrics across the vehicular mobility and the message communications networking dimensions of the overall system. We also study the design of a core network that employs Road Side Units (RSUs) that provide communications networking support in disseminating message flows to/from and among highway vehicles.

Cognitive impairment is a major concern for older adults especially in societies with increasing life expectancy and aging populations, since it reduces health related quality of life and increases caregiver burden. Possible and established factors for cognitive impairment not only include age, genetics, race/ethnicity, life style factors, but also metabolic syndrome, and evidence is accumulating that links environmental risk factors to cognitive impairment such as air pollution and noise exposure. As one of the risk factors of cognitive impairment, metabolic syndrome (MetS) refers to a collection of reversible pathophysiologic conditions including insulin resistance, obesity, dyslipidemia and hypertension. It is highly prevalent in the Hispanic population especially among those aged 60 years or older. In this dissertation, our aim was to investigate the influence of exposures to traffic-related air pollution and noise on incident metabolic syndrome and cognitive impairment, and whether the presence of metabolic syndrome would modify the association between air pollution or noise exposure and cognitive decline in elderly Mexican-Americans.

The following studies used data from the Sacramento Area Latino Study on Aging (SALSA), a prospective cohort study of 1789 Mexican-Americans aged 60-101, who were living in the Sacramento Area of California between 1998 and 2007. Based on participants’ residential addresses at baseline, we estimated local traffic-related nitrogen oxides (NOx) exposure using the California Line Source Dispersion Model version 4 (CALINE4), and traffic noise employing the SoundPLAN software package.

For 1,554 SALSA participants who were free of all five components of MetS at baseline according to the recommendations of the Third Adult Treatment Panel of the National

Cholesterol Education Program (NCEP ATP III), we investigated associations between modeled traffic-related NOx or noise pollution and incident metabolic syndrome or its components using Cox regression models with calendar time as the underlying time scale. We found that per unit increase in traffic-related NOx (2.29 parts per billion (ppb)) the hazard ratio (HR) for having low level of high-density lipoprotein cholesterol (HDL-cholesterol) increased by 15% (HR=1.15, 95% CI: 1.04–1.28), and for each 11.6 decibels (dB) increase in noise the risk of developing metabolic syndrome increased by 17% (HR=1.17, 95% CI: 1.01–1.35).

Some epidemiological studies started to focus on associations between air pollution and cognitive function recently, while the role of traffic noise in relation to cognitive impairment is under-studied. Here, we examined association between traffic-related noise pollution and dementia/ cognitive impairment without dementia (CIND) that developed newly over a 10-year follow-up period among 1,612 participants who were free of dementia/CIND at enrollment. Using Cox proportional hazard models, we observed that per 11.6dB (interquartile range, IQR) increase in 24-hour noise, the hazard of developing dementia/CIND increased (HR = 1.24, 95% CI: 1.00, 1.53) during follow-up; estimates were slightly lower (HR = 1.19, 95% CI: 0.95, 1.49) when adjusting for modeled local air pollution exposure from traffic sources. Overall, the risk of dementia/CIND was elevated when 24-hour and nighttime noise were higher than 75dB and 65dB, respectively.

In the third study, we investigated whether the presence of metabolic dysfunction (obesity, hyperglycemia and low HDL-cholesterol) modifies associations between air pollution or noise exposures and incident dementia or CIND. Among the 1,612 participants from SALSA study who were cognitively normal at the baseline, we used Cox proportional hazard models with calendar time as the underlying time scale to estimate the joint effects of air pollution and noise exposures and several metabolic dysfunctions, specially obesity, hyperglycemia, or low HDL-cholesterol. We found that the risk of developing dementia/CIND increased most (more than 2-fold) among SALSA participants who were exposed to high levels of traffic-related NOx (≥ 3.44 ppb [75th percentile]) (HR = 2.36, 95% CI = 1.41, 3.97) or 24-hour noise (≥ 65 dB) (HR = 2.21, 95% CI = 1.26, 3.89), respectively, and had hyperglycemia. The estimated hazard ratios for dementia/CIND were similarly increased with traffic related air pollution or noise exposures among participants with low HDL-cholesterol but no difference were seen for obesity.

Employing data from one of the large population-based studies of Mexican-Americans, in which repeated anthropometric measurements and sampling of biomarkers as well as repeated cognitive function testing, we added for the linkage between traffic-related air pollution and noise exposures with metabolic syndrome and cognitive impairment. We additionally improve the understanding for the role that metabolic dysfunctions play in the association between traffic-related exposures and cognitive decline. Early identification and treatment of people with metabolic dysfunction as well as prevention approaches that restricting the traffic-related exposures in residential neighborhoods might provide an effective avenue to generate public health benefits in vulnerable populations of elderly.

As quantum computing, communication, and sensing play increasingly important roles in the coming decades, and the demands for high sensitivity imaging grow rapidly, the sensitivity of photodetectors has become a main concern in several key technology sectors. This thesis reports two novel light detection and internal amplification mechanisms that enable photodiodes to achieve desired characteristics such as high detection efficiency, single photon sensitivity, low noise and high speed.

Cycling excitation process (CEP) is an intrinsic signal amplification mechanism that was firstly discovered in a heavily-doped compensated Si p/n junction device. By relaxing the k-selection rule (i.e. conservation of momentum), CEP detectors possess high gain, high efficiency, and ultralow noise. Above all, CEP detectors can be made from disordered materials such as amorphous silicon (a-Si), which allows low cost manufacturing and scalability to large array size for intended applications. An a-Si photodiode is demonstrated with ultra-high gain-bandwidth product of 2.25 THz and low noise, based on a very simple structure.

Dark current is a key challenge for photodiodes to detect extremely low power such as single photon as it can produce shot noise. A proper designed based on band gap engineering is shown to reduce the dark current of a-Si CEP detectors. The result of temperature dependent dark current measurement conveys the message that the dark current mechanism of a-Si CEP detectors is indirect tunneling followed by Poole-Frenkel effect. The key part is how to block the electron tunneling without affecting the device photo response. Cupric oxide (Cu2O) is an intrinsic p-type semiconductor material, with electron affinity of -3.2 eV and band gap of 2.1 eV. By inserting a thin layer of Cu2O between the a-Si and top electrode, the new structure has a much large electron tunneling barrier. On the other hand, the photogenerated holes can move as before without any additional hole blocking barrier. The measured results support that Cu2O based dark current reduction CEP devices have at least 1-2 orders of magnitude lower dark current at reverse bias 5 V compared to the typical a-Si CEP detectors while the photo responsivity remain the same.

Realizing that CEP effect occurs in a thin layer of a-Si, we designed an a-Si CEP detector without any semiconductor substrate. The key challenges are to improve the absorption efficiency and frequency response. Utilizing the localized surface plasmon resonance (LSPR) effect incorporated with CEP effect, a plasmonically enhanced a-Si detector achieves high external quantum efficiency with a record fast impulse response of 170 ps (FWHM). This approach raises the possibility of making detectors out of amorphous materials for high frame rate imaging and optical communications in spite of the low carrier mobilities in the materials.

The second optical signal amplification mechanism was observed in organometallic perovskite based detectors when the input power was down to a few or even single photon. It is a quasi-persistent photo response which takes tens of seconds for the current level to increase and reach to saturation after the absorption of a single photon. Based on the observation, we proposed an internal amplification mechanism, ionic impact ionization (I3), to elucidate the phenomenon, which involves a cascade process of ion migration.

Missing data occur frequently in surveys, clinical trials as well as other real data studies. In the analysis of incomplete data, one needs to correctly identify the missing mechanism and then adopt appropriate statistical procedures. Recently, the analysis of missing data has gained more and more attention. People start to investigate the missing data analysis in several different areas. This dissertation concerns two projects. First, we propose a Bayesian solution to data analysis with non-ignorable missingness. The other one is the non-parametric test of missing mechanism for incomplete multivariate data.

First, Bayesian methods are proposed to detect non-ignorable missing and eliminate potential bias in estimators when non-ignorable missing presents. Two hierarchical linear models, pattern mixture model and selection model, are applied to a real data example: the National Assessment of Education Progress (NAEP) education survey data. The results show that the Bayesian methods can correctly recognize the missingness mechanism and provide model-based estimators which can eliminate the possible bias due to non-ignorable missing. We also evaluate the goodness-of-fit of these two proposed models using two methods: the comparison of the real data with the predictive posterior distribution and the residual analysis by cross validation. A simulation study compares the performance of the two proposed Bayesian methods with the traditional design-based methods under different missing mechanisms and show the good properties of the Bayesian methods. Further, we discuss the three commonly used model selection criteria: the Bayes factor, the deviance information criterion and the minimum posterior predictive loss approach. Due to the complicated calculation of the Bayes factor and the uncertainty of the DIC, we conduct the last approach, which fails to correctly detect the real model structure for the hierarchical linear model.

Second, as an alternative to the fully specified model-based Bayesian method, a novel non-parametric test is proposed to detect the missing mechanism for multivariate missing data. The proposed test does not need any distributional assumptions and is proven to be consistent. A simulation study demonstrates that it has well controlled type I error and satisfactory power against a variety of alternative hypotheses.

Introduction

Methods

Results

Conclusions