Search

Feed efficiency, greenhouse gas (GHG) emissions, and animal productivities are major challenges facing the animal industry, especially the ruminant livestock industry. The rumen microbiome has attracted tremendous research interests because it is the major producer of the nutrients that can be assimilated by ruminants and contributor of methane production, and because it affects both feed efficiency and product (meat and milk) quality. In the past decade, meta-omic studies of the rumen microbiome produced huge amount of data about rumen microbiome and its association with feed efficiency, animal productivity, and methane emissions. The newly available data provide new opportunities to address the above challenges facing the ruminant livestock industry. However, the usage of such data was limited mainly because of their high dimensions. The developments of methods and algorithms of bioinformatics and machine learning (ML) provide new tools to analyze the rumen microbiome data beyond correlation analyses. Thus, in the studies presented in this dissertation, we explored ML methods coupled with bioinformatics analysis in investigating the quantitative relationship between relative sequence abundance of members of the rumen microbiome and several animal productivity traits and gain new insights into the roles of rumen microbiome in affecting ruminant production, with a focus on methane emissions from sheep and milk production traits of Holstein cows. The overall objectives of our studies were to investigate the quantitative relationship of members of the rumen microbiome with methane emissions and animal productivity and identify potential biomarkers and prediction models of the above animal traits. The abundance of some rumen microbes has been found to be associated with methane emissions from ruminants. We hypothesized that the growth rates, which are directly determined by metabolic activities and physiological features, of some rumen microbes might also be associated with methane emissions. In the first study, we tested the above hypothesis by estimating the growth rates of rumen bacteria that were represented by peak-to-trough ratio, which is the ratio between the ori region and the ter region, of metagenome-assembled genomes (MAGs) and examining the correlation between growth rate and methane emissions. The growth rates of 21 bacterial species (represented by MAGs) taxonomically classified to 12 genera in two groups of sheep differing in methane yield (high vs. low). Faecalibacterium prausnitzii was the only bacterial species with significantly different growth rates between the two groups of sheep. Examining the association between the relative abundance of rumen bacteria and methane emissions, we found 79 genera or species of rumen bacteria whose relative abundances differed between the high- and low-methane yield sheep. Lactate-producing bacteria (i.e., Intestinibaculum porci) and lactate-utilizing bacteria (i.e., Megasphaera elsdenii) were enriched in the rumen of the low-methane yield sheep. Additionally, we compared the expression of the genes (the transcripts) that were mapped to the genomes of F. prausnitzii using metatranscriptomic data. The expressions of the genes involved in energy metabolism of F. prausnitzii were upregulated, including those involved in butyrate production, in the rumen of the low-methane yield sheep. This finding suggests that butyrate production by F. prausnitzii might be promoted by lactate-metabolizing bacteria. Thus, the growth rate and expression of the genes involved in energy metabolism of F. prausnitzii may be useful biomarkers of methane emissions from sheep. The results of this study suggest that growth rates of individual rumen bacteria can be estimated from metagenomic sequence data, and growth rates of some rumen bacteria may affect methane emissions from ruminants. We also examined the correlation between methane emissions and the relative abundance of MAGs or gene expressions of some lactate-metabolizing bacteria. We revealed negative associations between the growth rate of F. prausnitzii in the rumen and methane yield in sheep and between the growth rate of F. prausnitzii and the expression of its genes involved in energy metabolism including butyrate production. We also proposed a working model to explain the promoted butyrate productions and enriched lactate-metabolizing bacteria in the rumen of low-methane yield sheep. In the second study, we developed new prediction models to improve prediction of methane emissions from sheep by including metataxonomic data of the rumen microbiota, in addition to animal-related data and rumen fermentation characteristics data. To overcome high-dimensionality of the metataxonomics data, we used supervised ML methods, including random forest combined with boosting (RF-B), least absolute shrinkage and selection operator (LASSO), generalized linear mixed model with LASSO (glmmLasso), and smoothly clipped absolute deviation (SCAD), to make variable selection and model fitting. We successfully developed one prediction model each, which, beside animal predictor variables, included genera (known genera such as Megasphaera, Oribacterium Selenomonas, Moryella, Prevotella_7, and unclassified candidate genera) of microbes as predictor variables, for prediction of methane production (amount of methane per animal per day) and methane yield (amount pf methane per kg of dry matter intake). Evaluation showed that both the animal-based prediction model and dry matter intake-based prediction model had considerable decreases in root mean square prediction error (RMSPE), mean absolute error (MAE), and Bayesian information criterion (BIC), and an increase in Lin’s concordance correlation coefficient (CCC) when compared to the models that only contained animal predictor variables. The new models also resulted in improved correlation between the observed and the predicted methane emissions. Our results demonstrated that the inclusion of rumen microbes as predictor variables, besides animal- and feed-related variables, could improve the accuracy and robustness of methane prediction models, and rumen microbiome data should be included in prediction models of methane emissions from ruminants. The metatranscriptomic analysis revealed that the gene expressions of Megasphaera, including those involved in propionate and butyrate production, were upregulated in low methane sheep. All the gene expressions of Oribacterium and Selenomonas showed slight tendency in two groups of sheep. In summary, the second aim revealed the inclusion of the relative abundance of some rumen microbes at genera-level using ML improved prediction of methane emissions. The important biomarkers selected by ML such as Megasphaera were interpreted based on upregulated expression of energy-related genes. Besides methane emissions, milk production can also be associated with rumen microbiome because it produces many of the precursors of milk production and affects feed utilization efficiency. Thus, a third study was dedicated to developing new models to predict milk production traits by including rumen microbes as predictor variables. We collated data from one study involving 334 lactating Holstein cows fed the same diet, which contained animal data, rumen fermentation characteristics data, and metataxonomic data of the rumen microbiota. We evaluated multiple ML methods in developing models to predict four milk production traits (milk yield, milk fat content, milk protein content, and milk lactose content). Amplicon sequence variants (ASVs) of bacteria as predictor variables resulted in better models than known genera or all genera of bacteria. For each milk production trait, we obtained 47 models by fitting the data using the ML methods (14 in total). The best prediction model for each milk production trait was selected based on evaluation of the prediction performance and fitting performance. The best performing models included animal variables, rumen fermentation characteristics variables, and ASVs as predictor variables. As evaluated based on RMSPE, MAE, CCC, Akaike information criterion, and adjusted R-squared, these new models improved prediction accuracy and reduced biases compared to the models that contained only animal variables or animals and rumen fermentation characteristics variables. The predictions of metagenome-functions ASV were then made to explore their functions in rumen. By comparing with ASV negatively associated with milk yield, the abundances of genes with functions related to biosynthesis of products such as amino acid, B vitamins, and lipopolysaccharide were more abundant in ASV positively associated with milk yield. Also, ASV positively associated with milk lactose had more genes related to starch and sucrose metabolism than that in ASV negatively associated with milk lactose. Analysis of the predicted functions of the ASVs included in the prediction models showed that the rumen microbiome may affect the host performance by affecting sugar metabolism and biosynthesis of amino acids and B vitamins. The overall results of this dissertation provide more insights into the biomarkers in rumen microbiome and their functions related to ruminant productions than was previously studied. Future research is warranted to explore the associations of identified bacteria, especially with respect to their growth, gene abundance and gene transcriptions, with methane emissions or milk productions from ruminants.

Methane emission from the rumen leads to approximately 10 % loss of the ingested energy, at the conversion of digestible to metabolizable energy. Besides research on mitigation of methane emission, much research interests have also been gravitated towards development of prediction models of methane emissions from livestock because the global warming is reducing agriculture productivity (Johnson and Johnson, 1995). An accurate prediction of enteric methane production from cattle and sheep can assist in balancing the increased livestock production with subsequent environmental impacts. Methane is an inevitable byproduct of the microbial fermentation processes in the rumen, and certain ruminal microbes have direct impacts to methane production (Morgavi et al., 2010). Thus, we hypothesized that the inclusion of individual microbial groups as predictor variables could improve the robustness and accuracy of prediction models. However, inclusion of microbial variables into prediction models can result in overfitting. Machine-learning algorithms can automatically select the key explanatory predictors, and Linear Mixed Models can provide a framework to predict random effects describing between animal variations. We proposed three novel frameworks for subset selections of microbial variables (MV) and one framework for generalized linear mixed models (GLMM) using L1-penalization (GLMMLASSO) selection with cross-validations (CV) to address the overfitting problems and to develop parsimonious prediction models of methane production.Methane emission can be expressed as unit methane per animal per d, per unit of dry matter intake (DMI), or per unit of metabolic body weight (MBW) per d. Thus, we developed prediction models based on g CH4/per animal/d (Animal-based models), g CH4/kg DMI (DMI-based models), and g CH4/kg metabolic bodyweight/d (MBW-based models). Two datasets were used: one dataset from a study that compared the rumen microbiota and methane production among sheep in New Zealand, while the other dataset that was collated from two datasets (one generated using dairy cows in Finland, and the other generated using steers in Australia). The cattle datasets were generated from studies that used different anti-methane feed additives to mitigate methane emission. Each dataset contained both animal data and relative sequence abundance (RSA) of genera of rumen microbes including bacteria, methanogens, protozoa, and fungi. Relative abundance of a genus was expressed as % of the sequences assigned to that genus over the total sequences of a marker gene, 16S rRNA gene for bacteria and methanogens, 18S rRNA gene for protozoa, and ITS1 for fungi). Subset and GLMMLASSO selections of MV combined with CV were based on minimal Bayesian information criterion (BIC). Linear mixed effects models were built based on the MV selected. The cross-validation was used to identify the best subset of MV that resulted in the lowest mean square prediction error (MSPE) to include in the prediction models. We also compared our new models with traditional models that only contained DMI, acetate: propionate ratio, and BW.From the sheep dataset, we developed 3 parsimonious models when the size of the pool of potential variables was limited to =132, but parsimonious models were not developed when the size of the pool of potential variables was over 300. Most importantly, GLMMLASSO successfully converged when the penalty parameter that controls the shrinkage, Lambda, was set between 0 to 1,000, selecting about 10 variables for all the models (animal-, DMI-, or MBW-based) when the size of pool of potential predictor variables was limited to be between 132 and 310. The GLMMLASSO approach combined with CV identified the important variables to explain the effects of the 8 feed additives in the combined cow dataset. The random effect associated with the between-animal variance component was small and of similar magnitude to the random error variance component. The linear mixed effects models based on GLMMLASSO selection of MV have root mean square error as the percentage of the mean of methane emission (RMSPE %) reduced by 2.3 percent points and performed better than those based on forward-stepwise MV selection and the traditional models based only on animal variables. Log transformation of the microbial data generally improved model performance, probably due to a better monotonicity of the MV. This thesis research indicates that individual groups of rumen microbes can be included in methane prediction models to improve prediction of methane production from sheep and cows.In conclusion, we built frameworks for model selections using MV, and GLMMLASSO combined with CV and forward-stepwise selection allowed identification of significant MV that can be included in methane prediction models that solved the overfitting problem and improved the prediction accuracy. GLMMLASSO selection coupled with CV is a useful method to extract a parsimonious and significant subset of MV from hundreds of MV that can improve the accuracy of methane prediction models. This is the first research to develop methane prediction models that contain rumen microbes as predictor variables. Future research can focus on the exploration of the common microbial predictor variables in the cow and the sheep datasets to understand the contribution of the microbes to methane production in the rumen and to develop models with a more mechanistic nature.

This dissertation studies monetary policy in emerging market economies and addresses three important questions from both a normative and positive perspective. Chapter 1 studies how central banks should react when capital flows are volatile. The analytical framework is a Markov-switching dynamic stochastic general equilibrium model that features time-varying external volatility. Computational results suggest that central banks can improve welfare and maintain macroeconomic stability if they allow the response coefficients in the interest rate rule to vary according to the external volatility regime. The optimal simple rule suggests that central banks should target inflation when external volatility is low and stabilize exchange rates when it is high, which is akin to the “leaning against the wind” approach adopted by some emerging market central banks. Chapter 2 studies the optimal inflation target for an emerging market central bank. Existing research on advanced economies shows that targeting core inflation enables monetary policy to maximize welfare. This result is examined in the context of emerging market economies, where a large proportion of households are credit constrained and the share of food expenditures in total consumption expenditures is high. Results obtained using an open economy model with incomplete financial markets indicate that headline inflation targeting improves welfare outcomes. The optimal price index includes a positive weight on food prices but assigns zero weight to import prices. Chapter 3 studies the inter-sectoral distributional effects of monetary policy in emerging market economies. Emerging market economies with fast-growing tradable sectors often face appreciation pressure, and they tend to use monetary policy to postpone currency appreciation and maintain export competitiveness. A two-sector, heterogeneous-agent model with incomplete financial markets is developed to study the inter-sectoral distributional effects of such policy choices. Relative to inflation targeting, exchange rate management can temporarily benefit households in the tradable sector in high-growth periods, but these households are worse off under the welfare criterion due to higher future consumption volatility. Capital controls and fixed exchange rate regimes amplify those distributional effects.

As Moore's law continues to 20nm and below, traditional CMOS device faces severe short channel effects. Industry is switching from traditional CMOS to FinFET in order to keep Moore's law alive. Due to the three-dimensional structure of FinFET, many challenges need to be solved. After that, FinFET will finally be able to replace traditional CMOS in the semiconductor industry. This thesis discusses the manufacturing challenges of FinFET. In addressing these challenges, characterization of the FinFET standard cells has been done. The characterization is based on saturation current, leakage current, implantation angle and the average edge placement error at metal one layer. Three design variables, including the metal pitch, the fin pitch and the fin width are optimized to achieve better design quality. Standard cell library which contains combinatorial cells as well as sequential cells are characterized and optimized. Two optimization scenarios are included in the final results. One is performance driven, optimizing the saturation current and the leakage current, while the other is manufacturability driven, optimizing the implantation angle and the average EPE. The optimization results show the tradeoff between performance and manufacturability.