Search

Experiments that longitudinally collect RNA sequencing (RNA-seq) data can provide transformative insights in biology research by revealing the dynamic patterns of genes. Such experiments create a great demand for new analytic approaches to identify differentially expressed (DE) genes based on large-scale time-course count data. Existing methods, however, are suboptimal with respect to power and may lack theoretical justification. Furthermore, most existing tests are designed to distinguish among conditions based on overall differential patterns across time, though in practice, a variety of composite hypotheses are of more scientific interest. Finally, some current methods may fail to control the false discovery rate. In this paper, we propose a new model and testing procedure to address the above issues simultaneously. Specifically, conditional on a latent Gaussian mixture with evolving means, we model the data by negative binomial distributions. Motivated by Storey (2007) and Hwang and Liu (2010), we introduce a general testing framework based on the proposed model and show that the proposed test enjoys the optimality property of maximum average power. The test allows not only identification of traditional DE genes but also testing of a variety of composite hypotheses of biological interest. We establish the identifiability of the proposed model, implement the proposed method via efficient algorithms, and demonstrate its good performance via simulation studies. The procedure reveals interesting biological insights, when applied to data from an experiment that examines the effect of varying light environments on the fundamental physiology of the marine diatom Phaeodactylum tricornutum.

We consider small area estimation for the departure times of recreational anglers along the Atlantic and Gulf coasts of the United States. A Bayesian area-level Fay–Herriot model is considered to obtain estimates of the departure time distribution functions. The departure distribution functions are modelled as circular distributions plus area-specific errors. The circular distributions are modelled as projected normal, and a regression model is specified to borrow information across domains. Estimation is conducted through the use of a Hamiltonian Monte Carlo sampler and a projective approach onto the probability simplex. The Canadian Journal of Statistics 44: 416–430; 2016 © 2016 Statistical Society of Canada

Small-angle X-ray scattering (SAXS) is a technique for obtaining low-resolution structural information about biological macromolecules, by exposing a dilute solution to a high-intensity X-ray beam and capturing the resulting scattering pattern on a two-dimensional detector. The two-dimensional pattern is reduced to a one-dimensional curve through radial averaging; that is, by averaging across annuli on the detector plane. Subsequent analysis of structure relies on these one-dimensional data. This paper reviews the technique of SAXS and investigates autocorrelation structure in the detector plane and in the radial averages. Across a range of experimental conditions and molecular types, spatial autocorrelation in the detector plane is present and is well-described by a stationary kernel convolution model. The corresponding autocorrelation structure for the radial averages is non-stationary. Implications of the autocorrelation structure for inference about macromolecular structure are discussed.

Many environmental data sets have a continuous domain, in time and/or space, and complex features that may be poorly modelled with a stationary (in space and time) Gaussian process (GP). We adapt stochastic volatility modelling to this context, resulting in a stochastic heteroscedastic process (SHP), which is unconditionally stationary and non-Gaussian. Conditional on a latent GP, the SHP is a heteroscedastic GP with non-stationary (in space and time) covariance structure. The realizations from SHP are versatile and can represent spatial inhomogeneities. The unconditional correlation functions of SHP form a rich isotropic class that can allow for a smoothed nugget effect. We apply an importance sampling strategy to implement pseudo maximum likelihood parameter estimation for the SHP. To predict the process at unobserved locations, we develop a plug-in best predictor. We extend the single-realization SHP model to handle replicates across time of SHP realizations in space. Empirical results with simulated data show that SHP is nearly as efficient as a stationary GP in out-of-sample prediction when the true process is a stationary GP, and outperforms a stationary GP substantially when the true process is SHP. The SHP methodology is applied to enhanced vegetation index data and US NO3 deposition data for illustration.

Process-based model analyses are often used to estimate changes in soil organic carbon (SOC), particularly at regional to continental scales. However, uncertainties are rarely evaluated, and so it is difficult to determine how much confidence can be placed in the results. Our objective was to quantify uncertainties across multiple scales in a process-based model analysis, and provide 95% confidence intervals for the estimates. Specifically, we used the Century ecosystem model to estimate changes in SOC stocks for US croplands during the 1990s, addressing uncertainties in model inputs, structure and scaling of results from point locations to regions and the entire country. Overall, SOC stocks increased in US croplands by 14.6 Tg C yr -1 from 1990 to 1995 and 17.5 Tg C yr -1 during 1995 to 2000, and uncertainties were ± 22% and ± 16% for the two time periods, respectively. Uncertainties were inversely related to spatial scale, with median uncertainties at the regional scale estimated at ± 118% and ± 114% during the early and latter part of 1990s, and even higher at the site scale with estimates at ± 739% and ± 674% for the time periods, respectively. This relationship appeared to be driven by the amount of the SOC stock change; changes in stocks that exceeded 200 Gg C yr -1 represented a threshold where uncertainties were always lower than ±100%. Consequently, the amount of uncertainty in estimates derived from process-based models will partly depend on the level of SOC accumulation or loss. In general, the majority of uncertainty was associated with model structure in this application, and so attaining higher levels of precision in the estimates will largely depend on improving the model algorithms and parameterization, as well as increasing the number of measurement sites used to evaluate the structural uncertainty.

Ground-based data on crop production in the USA is provided through surveys conducted by the National Agricultural Statistics Service (NASS) and the Census of Agriculture (AgCensus). Statistics from these surveys are widely used in economic analyses, policy design, and for other purposes. However, missing data in the surveys presents limitations for research that requires comprehensive data for spatial analyses. We created comprehensive county-level databases for nine major crops of the USA for a 16-yr period, by filling the gaps in existing data reported by NASS and AgCensus. We used a combination of regression analyses with data reported by NASS and the AgCensus and linear mixed-effect models incorporating county-level environmental, management, and economic variables pertaining to different agroecozones. Predicted yield and crop area were very close to the data reported by NASS, within 10% relative error. The linear mixed-effect model approach gave the best results in filling 84% of the total gaps in yields and 83% of the gaps in crop areas of all the crops. Regression analyses with AgCensus data filled 16% of the gaps in yields and crop areas of the major crops reported by NASS.

Models are central to global change analyses, but they are often parameterized using data that represent only a portion of heterogeneity in a region. This creates uncertainty in the results and constrains the reliability of model inferences. Our objective was to evaluate the uncertainty associated with differential scaling of parameterization data to model soil organic carbon stock changes as a function of US agricultural land use and management. Specifically, we compared analyses in which model parameters were derived from field experimental data that were scaled to the entire US vs. the same data scaled to climate regions within the country. We evaluated the effect of differential scaling on both bias and variance in model results. Model results had less variance by scaling data to the entire country because of a larger sample size for deriving individual parameter values, although there was a relatively large bias associated with this parameterization, estimated at 2.7TgCyr � 1 . Even with the large bias, resulting confidence intervals from the two parameterizations had considerable overlap for the estimated national rate of SOC change (i.e. 77% overlap in those intervals). Consequently, the results were relatively similar when focusing on the uncertainty rather than solely on the mean estimate. In contrast, large biases created less overlap in confidence intervals for the change rates within individual climate regions, compared with the national estimates. For example, the overlap in resulting intervals from the two parameterizations was only 32% for the warm temperate moist region, with a corresponding bias of 3.1TgCyr � 1 . These findings demonstrate that there is a greater risk of making erroneous inferences because of large biases if models are parameterized with broader scale information, such as an entire country, and then used to address impacts at a finer spatial scale, such as subregions within a country. In addition, the study demonstrates a trade-off between variance and bias in model results that depends on the scaling of data for model parameterization.

No-tillage (NT) management has been promoted as a practice capable of offsetting greenhouse gas (GHG) emissions because of its ability to sequester carbon in soils. However, true mitigation is only possible if the overall impact of NT adoption reduces the net global warming potential (GWP) determined by fluxes of the three major biogenic GHGs (i.e. CO2, N2O, and CH4). We compiled all available data of soil-derived GHG emission comparisons between conventional tilled (CT) and NT systems for humid and dry temperate climates. Newly converted NT systems increase GWP relative to CT practices, in both humid and dry climate regimes, and longer-term adoption (>10 years) only significantly reduces GWP in humid climates. Mean cumulative GWP over a 20-year period is also reduced under continuous NT in dry areas, but with a high degree of uncertainty. Emissions of N2O drive much of the trend in net GWP, suggesting improved nitrogen management is essential to realize the full benefit from carbon storage in the soil for purposes of global warming mitigation. Our results indicate a strong time dependency in the GHG mitigation potential of NT agriculture, demonstrating that GHG mitigation by adoption of NT is much more variable and complex than previously considered, and policy plans to reduce global warming through this land management practice need further scrutiny to ensure success.

Uncertainty was quantified for an inventory estimating change in soil organic carbon (SOC) storage resulting from modifications in land use and management across US agricultural lands between 1982 and 1997. This inventory was conducted using a modified version of a carbon (C) accounting method developed by the Intergovernmental Panel on Climate Change (IPCC). Probability density functions (PDFs) were derived for each input to the IPCC model, including reference SOC stocks, land use/management activity data, and management factors. Change in C storage was estimated using a Monte-Carlo approach with 50 000 iterations, by randomly selecting values from the PDFs after accounting for dependencies in the model inputs. Over the inventory period, mineral soils had a net gain of 10.8 Tg C yr−1, with a 95% confidence interval ranging from 6.5 to 15.3 Tg C yr−1. Most of this gain was due to setting-aside lands in the Conservation Reserve Program. In contrast, managed organic soils lost 9.4 Tg C yr−1, with a 95% confidence interval ranging from 6.4 to 13.3 Tg C yr−1. Combining these gains and losses in SOC, US agricultural soils accrued 1.3 Tg C yr−1 due to land use and management change, with a 95% confidence interval ranging from a loss of 4.4 Tg C yr−1 to a gain of 6.9 Tg C yr−1. Most of the uncertainty was attributed to management factors for tillage, land use change between cultivated and uncultivated conditions, and C loss rates from managed organic soils. Based on the uncertainty, we are not able to conclude with 95% confidence that change in US agricultural land use and management between 1982 and 1997 created a net C sink for atmospheric CO2.

A new sampling-based Bayesian approach for fractionally integrated autoregressive moving average (ARFIMA) processes is presented. A particular type of ARMA process is used as an approximation for the ARFIMA in a Metropolis–Hastings algorithm, and then importance sampling is used to adjust for the approximation error. This algorithm is relatively time-efficient because of fast convergence in the sampling procedures and fewer computations than competitors. Its frequentist properties are investigated through a simulation study. The performance of the posterior means is quite comparable to that of the maximum likelihood estimators for small samples, but the algorithm can be extended easily to a variety of related processes, including ARFIMA plus short-memory noise. The methodology is illustrated using the Nile River data. Copyright © 2003 John Wiley & Sons, Ltd.

A number of recent empirical studies have generated skepticism about the effectiveness of generic advertising for beef. One of these studies, Jensen and Schroeter (1992), examines data collected from a panel of households in a carefully designed experimental test of television advertising. The present paper undertakes a reexamination of the Jensen and Schroeter data with two significant improvements in method. First, the analysis disaggregates beef purchases into three product types (ground beef steaks and roasts) and assesses advertising's impact on the demand for each type separately. Second, the present analysis uses an improved econometric method: Bayesian inference in a random effects Tobit model. Inference is based on simulations of a posterior distribution using Gibbs sampling and data augmentation. As far as advertising's effects are concerned, the results of this analysis reaffirm the Jensen and Schroeter finding: The experimental television advertising campaign was not effective in increasing household purchases of beef. Plusieurs etudes empiriques recentes ont souleve le doute quant a l'efficacite de la publicite generique sur le bœuf. Celle de Jensen et Schroeder (JS), notamment, portait sur les donnees recueillies aupres d'un groupe de menages dans le cadre d'une experience bien concue sur la publicitea la television. Les auteurs reexaminent les donnees de cette etude en apportant deux ameliorations majeures a la methode utilisee. En premier lieu, leur analyse desagrege les achats de bœuf en trois (viande hachee, biftecks et rotis) et evalue l'impact de la publicite sur la demande de chaque produit. Deuxiemement, l'analyse repose sur une meilleure methode econometrique, a savoir l'inference bayesienne dans un modele Tobit a effets aleatoires. L'inference s'appuie sur la simulation d'une distribution posterieure par echantillonnage de Gibb et enrichissement des donnees. En ce qui concerne l'incidence de la publicite, cette nouvelle analyse confirme les constatations de l'etude JS, soit que la campagne experimental de publicitea la television n'entraine pas une hausse des achats de bœuf par les menages.

We consider bootstrap construction and calibration of prediction intervals for nonGaussian autoregressions. In particular, we address the question of prediction conditioned on the last p observations of the process, for which we offer an exact simulation technique and an approximate bootstrap approach. In simulations for a variety of first-order autoregressions, we compare various nonparametric prediction intervals and find that calibration gives reasonably narrow prediction intervals with the lowest coverage probability mean squared error among the methods used.

The US National Resources Inventory (NRI) is a longitudinal panel survey conducted by the US Department of Agriculture's Natural Resources Conservation Service (formerly the Soil Conservation Service) in cooperation with the Iowa State University Statistical Laboratory. In its current form, the NRI is designed to assess conditions and trends at five year intervals for soil, water, and related natural resources on nonfederal lands of the US. The survey has evolved over a period of several decades, with the specific goal of providing information to support agricultural and environmental policy development and program implementation.