Showing papers in "Journal of Agricultural Biological and Environmental Statistics in 2000"
TL;DR: This paper uses covariates and an indication of sampling effort in an autologistic model to improve predictions of probability of presence for lattice data and adopts a Bayesian set-up and develops a Gibbs sampling estimation procedure.
Abstract: In this paper, we use covariates and an indication of sampling effort in an autologistic model to improve predictions of probability of presence for lattice data. The model is applied to sampled data where only a small proportion of the available sites have been observed. We adopt a Bayesian set-up and develop a Gibbs sampling estimation procedure. In four examples based on simulated data, we show that the autologistic model with covariates improves predictions compared with the simple logistic regression model and the basic autologistic model (without covariates). Software to implement the methodology is available at no cost from StatLib.
133 citations
TL;DR: In this paper, the authors discuss the extent to which a Mantel type of analysis between two proximity matrices agrees with Pearson's correlation analysis when both methods are applicable (i.e., the raw data used to calculate proximities are available).
Abstract: The space-time clustering procedure of Mantel was originally designed to relate a matrix of spatial distance measures and a matrix of temporal distance measures in a generalized regression approach. The procedure, known as the Mantel test in the biological and environmental sciences, includes any analysis relating two distance matrices or, more generally, two proximity matrices. In this paper, we discuss the extent to which a Mantel type of analysis between two proximity matrices agrees with Pearson's correlation analysis when both methods are applicable (i.e., the raw data used to calculate proximities are available). First, we demonstrate that the Mantel test and Pearson's correlation analysis should lead to a similar decision regarding their respective null hypothesis when squared Euclidean distances are used in the Mantel test and the raw bivariate data are normally distributed. Then we use fish and zooplankton biomass data from Lake Erie (North American Great Lakes) to show that Pearson's correlation statistic may be nonsignificant while the Mantel statistic calculated on nonsquared Euclidean distances is significant. After small-size artificial examples, seven bivariate distributional models are tried to simulate data reproducing the difference between analyses, among which three do reproduce it. These results and some extensions are discussed. In conclusion, particular attention must be paid whenever relations established between proximities are backtransposed to raw data, especially when these may display patterns described in the body of this paper.
125 citations
TL;DR: In this article, the authors present a practical guide to analysis of BACI studies when response variables are counts, including two commonly used analyses and one less common, but more appropriate, analysis.
Abstract: Before-after control-impact (BACI) studies are common observational studies conducted to determine environmental impacts of accidents or potential disturbances. In this paper, we present a practical guide to analysis of BACI studies when response variables are counts. Two commonly used analyses and one less common, but more appropriate, analysis are covered. The two common analyses fundamentally compare differences of differences, one using original units, the other using log-transformed units. The third analysis, which is less common, consists of estimating interaction effects in a quasi-likelihood generalized linear model with correlated errors (i.e., a generalized linear mixed model). We conclude that the two common analyses are of marginal utility when analyzing count data due to questions regarding interpretation of parameter estimates and treatment of zeros. These questions do not arise under the quasi-likelihood generalized linear model method, and it is the recommended approach. We illustrate the three techniques by analyzing data similar to that collected by an observational study of seabird counts on oiled and unoiled sites before and after the Exxon Valdez oil spill. Example data and SAS(r) code to conduct the three analyses are given.
119 citations
TL;DR: In this paper, the authors reviewed some common sampling designs from the perspective of their appropriateness for accurately estimating the total abundance of a rare population and performed a small simulation study in which the designs for three populations with varying numbers of rare elements were studied and provided information on the efficiency of the designs and the shapes of the sampling distributions of the estimators.
Abstract: Some common sampling designs are reviewed from the perspective of their appropriateness for accurately estimating the total abundance of a rare population. In addition, a small simulation study was performed in which the designs for three populations with varying numbers of rare elements were studied. The simulation study provided information on the efficiency of the designs and the shapes of the sampling distributions of the estimators. The best design was stratified random sampling in which strata were constructed so that quadrats with rare elements were confined to a single small stratum and that stratum was disproportionately oversampled. The estimator for this design had the smallest variance and a sampling distribution most similar to a normal distribution. Systematic sampling is a good second choice if an auxiliary variable on which to stratify is not available. The estimator had reasonably small variance, which was further reduced if adaptive cluster sampling was added. The disadvantages are that an estimate of the variance is not available if only a single systematic sample is taken and the sampling distribution of the estimator is discontinuous and definitely not shaped liked a bell curve.
59 citations
TL;DR: The authors prudently phrase their interpretation in terms of a weight of evidence argument rather than the much more difficult one of causation, encouraging scientists to use more complex models to realistically describe count data through time.
Abstract: McDonald, Erickson, and McDonald (2000) and Murtaugh (2000) have presented two interesting views of analysis and interpretation of data from before-after control-impact (BACI) studies. In my opinion, the general message they convey to the reader is that analysis of data from a BACI design, in combination with professional expertise, can add useful evidence as long as one is prudent about the resulting interpretations. Whether scientists who analyze BACI data will indeed use caution in interpretation, particularly in making statements regarding cause and effect, remains to be seen. McDonald et al.'s (2000) study is application specific, taking the reader through an analysis of bird count data from the Exxon Valdez oil spill (EVOS). The authors prudently phrase their interpretation in terms of a weight of evidence argument rather than the much more difficult one of causation. I see the primary usefulness of this paper as that of encouraging scientists to use more complex models to realistically describe count data through time. (Using these models in data analysis has been considerably enhanced by the increasing availability of various types of computer software.) Particularly with biological data, a conventional first step is to try an additive model on untransformed data or an additive model on log-transformed data with the hope that the log transform restores the assumptions of variance homogeneity and normally distributed error terms. The presence of data recorded on the same units through time necessitates that the Huynh-Feldt condition (also known as sphericity) holds in order to use a conventional repeated measures approach. In case this sphericity condition does not hold, there are references to make conservative adjustments to the usual F-tests. (I do wonder just how many people would take the trouble to test these assumptions and do the proper corrections if such corrections were not already built into the
55 citations
TL;DR: Mark-recapture studies of pinnipeds should attempt to use permanent marks in combination with tags to assess unbiased estimates of tag retention, and it is found that ignoring within-individual association of tag loss led to a bias in estimated tag retention.
Abstract: Mark-recapture studies of pinnipeds commonly use double-tagging to reduce bias of parameter estimates and to allow estimation of tag retention rates. However, most tag retention estimates assume independence of tag loss. Here we were able to identify when individual New Zealand fur seal (Arctocephalus forsteri) pups had lost both tags ; therefore, we tested the assumption of no association between the tag-loss rates of left and right tags. We also tested for differences in tag retention among three different types of plastic tag (Allflex® cattle, mini and button tags), between two attachment types (i.e., fixed or swivel), and whether retention varied among years and colonies sampled. We found strong evidence of within-individual tag loss association for most tags in most years, but little evidence that this varied among colonies. We found that ignoring within-individual association of tag loss led to a bias in estimated tag retention of 7.4-10.1%. Smaller rocks and greater crevice and ledge densities in colonies were associated with lower probabilities of tag retention. We suggest researchers should attempt to use permanent marks in combination with tags to assess unbiased estimates of tag retention.
50 citations
TL;DR: This paper analyzes two insect-related sets of agricultural field data that concern carabids in cereals, the incidence and spread of an aphid-vectored virus disease of lupins, and shows considerable aggregation for both the beetles and the infected lupin plants.
Abstract: This paper analyzes two insect-related sets of agricultural field data. Both comprise spatially referenced count data sampled on a series of occasions. One concerns carabids (ground beetles) in cereals, the other the incidence and spread of an aphid-vectored virus disease of lupins. For both sets, the major objective was to describe and quantify the stability through time of the spatial patterns found for each occasion; this was measured by the spatial association between successive samples. Traditional methods for analyzing count data focus on properties of the frequency distribution of the counts and use little or none of the spatial information in the sample. We used methods that utilized all the spatial information and that, by conditioning on the observed data, provided complementary inferences to the other methods. Our analyses are based on a class of methods termed spatial analysis by distance indices (SADIE). These methods provide indices and formal randomization tests, both for the spatial pattern in a single population and for the spatial association when the patterns of two populations are compared. Our analyses showed considerable aggregation for both the beetles and the infected lupin plants. Furthermore, both populations displayed positive association between successive samples that declined as the temporal lag increased. The beetles were affected greatly by the harvest of the cereal crop. The lupin infections showed maximal association for a 1-week lag despite the fact that the latent period of the virus was a fortnight; it was inferred that the observed pattern of new infections was probably tracking the pattern of the aphid vectors 2 weeks previously.
43 citations
TL;DR: In this paper, a Bayesian analysis of data under the mixed linear model is presented, where all of the parameters, including the variance components, are treated as random and the joint posterior distribution of all the parameters given the data is found.
Abstract: The mixed linear model is commonly used in animal breeding to predict or estimate the breeding values of individual animals. Breeding values are used to select animals for use in breeding subsequent generations. A traditional analysis of data under this model involves estimating the variance components with restricted maximum likelihood. These estimates are then used to find best linear unbiased predictors for the animal breeding values. A Bayesian analysis of data under this model involves treating all of the parameters, including the variance components, as random and finding the joint posterior distribution of all of the parameters given the data. Because the selection decision depends on the values of the variance components, a Bayesian analysis can yield different selection outcomes than the traditional analysis. We demonstrate both types of data analysis on data from an animal breeding experiment and compare the resulting selections.
35 citations
TL;DR: In this article, a Gibbs sampler was used to traverse the model space and predict chlorophyll concentrations in Lake Okeechobee using Bayesian model averaging (BMA) over the sampled models.
Abstract: Long-term eutrophication data along with water quality measurements (total phosphorous and total nitrogen) and other physical environmental factors such as lake level (stage), water temperature, wind speed, and direction were used to develop a model to predict chlorophyll a concentrations in Lake Okeechobee The semiparametric model included each of the potential explanatory variables as linear predictors, regression spline predictors, or product spline interactions allowing for nonlinear relationships A Gibbs sampler was used to traverse the model space Predictions that incorporate uncertainty about inclusion of variables and their functional forms were obtained using Bayesian model averaging (BMA) over the sampled models Semiparametric regression with Bayesian model averaging and spline interactions provides a flexible framework for addressing the problems of nonlinearity and counterintuitive total phosphorus function estimates identified in previous statistical models The use of regression splines allows nonlinear effects to be manifest, while their extension allows inclusion of interactions for which the mathematical form cannot be specified a priori Prediction intervals under BMA provided better coverage for new observations than confidence intervals for ordinary least squares models obtained using backwards selection Also, BMA was more efficient than ordinary least squares in terms of predictive mean squared error for overall lake predictions
33 citations
TL;DR: Two approaches, one based on the likelihood-ratio statistic and the other based on unconditioning Fisher's exact test, are examined for obtaining a p-value in the comparison of the combination of arthropod species present on a mystery carcass to the observed frequency distribution of species combinations on carcasses exposed to the elements for a month.
Abstract: Two approaches, one based on the likelihood-ratio statistic and the other based on unconditioning Fisher's exact test, are examined for obtaining a p-value in the comparison of the combination of arthropod species present on a mystery carcass to the observed frequency distribution of species combinations on carcasses exposed to the elements for a
32 citations
TL;DR: This article developed a two-stage analysis method that corrects for serial correlation and uses half-series means to assess temporal variation in the trajectories of the response through time in the Hubbard Brook Experimental Forest.
Abstract: The paired watershed experiments of Likens and coworkers in the Hubbard Brook Experimental Forest are examples of a classical design in ecology, in which a response in a manipulated unit is compared both to the response in the same unit before manipulation and to the response in an adjacent reference unit that remains undisturbed. Early proponents of this design did not attempt statistical analysis of their results but, more recently, before-after-control-impact analysis and randomized intervention analysis have been used by ecologists to draw statistical inferences from such data. These methods are simply two-sample comparisons (before vs. after) of between-unit differences, with significant results often interpreted as evidence for an effect of the intervention. This approach ignores variation caused by differences between units in the trajectories of the response through time, and it does not take into account possible serial correlation of errors. Consequently, the null hypothesis may be rejected much too often. I develop a new, two-stage analysis method that addresses these shortcomings by correcting for serial correlation and using half-series means to assess temporal variation. Unlike paired intervention analysis, the resulting test has close to the nominal level when the time course of the response is allowed to vary between units, but its power is extremely limited due to the lack of true replication in the design.
TL;DR: In this paper, a test of additivity in a combination of drugs/chemicals based on the interaction index was proposed for data resulting from an experiment involving single-agent exposure groups and s combination groups of interest.
Abstract: In this paper, we propose a test of additivity in a combination of drugs/chemicals based on the interaction index. The test is developed for data resulting from an experiment involving single-agent exposure groups and s combination groups of interest. The testing procedure begins with an overall size a test of additivity with s degrees of freedom. When this test is rejected, single degree-of-freedom tests combined with post hoc corrections for multiple testing are developed to determine which dose combinations are associated with departure from additivity. The method is illustrated using data associated with a patent application for a two-drug combination.
TL;DR: In this paper, a probit model with spatial correlation is applied to data from a field experiment, which characterizes the impact of management variables on potato leafroll virus net necrosis in potato tubers.
Abstract: A probit model with spatial correlation is applied to data from a field experiment, which characterizes the impact of management variables on potato leafroll virus net necrosis in potato tubers. In the estimation, each field plot is assigned distinct spatial autoregressive coefficients for the dependent variable and the residual to be estimated simultaneously with coefficients of the management variables. Statistical findings demonstrate that spatial correlation exists and varies across field plots. We also find that ignoring spatial correlation by plot results in inconsistent parameter estimates and leads to management strategies promoting overuse of insecticides. In contrast, incorporating spatial correlation by plot into the probit model yields empirical estimates that are consistent with past research and promotes more efficient insecticide use from both an individual and environmental perspective.
TL;DR: In this article, a state-space model is used to predict the spatial-temporal abundance of a tagged cohort of coho salmon in the wild, based on release-recovery and fishing effort data from several cohorts of a hatchery-reared coho stock originating from Washington state.
Abstract: Using fishery recoveries from a tagged cohort of coho salmon, the ocean spatial-temporal abundance of the cohort is predicted using a state-space model. The model parameters, which reflect spatial distribution, mortality, and movement, vary considerably between different cohorts. To evaluate the effect of proposed management plans on a future cohort, uncertainty in the cohort-specific parameters is accounted for by a hierarchic model. As an application, release-recovery and fishing effort data from several cohorts of a hatchery-reared coho salmon stock originating from Washington state are used to calculate maximum likelihood estimates of the hyperparameters. Markov chain Monte Carlo is used to approximate the likelihood for the hyperparameters. The Markov chain simulates the sampling distribution of the state-space model parameters conditional on the data and the estimated hyperparameters and provides empirical Bayes estimates as a by-product. Given the estimated hyperparameters and the hierarchic model, fishery managers can simulate the variation in cohort-specific parameters and variation in the migration and harvest processes to more realistically describe uncertainty in the results of any proposed management plan.
TL;DR: This work considers the problem of combining temporally correlated environmental data from two measurement systems and presents two methods for combining the data and for using the combined data to detect trend, which are both currently in use in the western United States.
Abstract: We consider the problem of combining temporally correlated environmental data from two measurement systems. More specifically, we suppose that an environmental variable has been measured at regular intervals for a relatively long period of time using one measurement system and that a newer, possibly cheaper or more reliable measurement system has been in operation in tandem with the old system for a relatively short period of time. We suppose that, for purposes of detecting changes or trends in the variable over time, the time series corresponding to the new system only is too short so that it is desirable to somehow combine it with the longer time series from the old system. We present two methods for combining the data and for using the combined data to detect trend. The first is a frequentist analysis of an autoregressive moving-average time series model featuring a common time trend, measurement errors with system-specific biases and variances, and missing data. The second method is a Bayesian analysis of a similar model that is implemented by a Markov chain Monte Carlo procedure. We use the methodology to combine and analyze snow water equivalent data from manual snow surveys (an old measurement system) and snow telemetry (a newer system), which are both currently in use in the western United States.
TL;DR: In this article, the authors proposed methods of risk assessment that exploit covariate information for making specific recommendations for crop yield risk assessment using data from a multienvironment trial on triticale (Triticosecale WITTMARK).
Abstract: Yield risk is an important criterion for crop cultivar choice. Two kinds of risk may be of interest, (1) the probability that the yield of a cultivar falls short of some critical level and (2) the probability of a cultivar being outperformed by another cultivar. These risks are governed by the average cultivar performances and by the variability of performances across diverse environments. As a result of among-environment variation, yield in a new environment is, to a large degree, unpredictable. Predictability may be improved by taking into account covariate information on the environments. The present paper proposes methods of risk assessment that exploit covariate information for making specific recommendations. Exact confidence limits for estimated risks are provided. The methods are exemplified using data from a multienvironment trial on triticale (Triticosecale WITTMARK).
TL;DR: In this paper, the influence of design characteristics on the statistical inference for an ecotoxicological hazard-based model using simulated survival data is described, and a comparison of the coverage probabilities for confidence limits arising from the profile-likelihood approach and the Wald-based approach is carried out.
Abstract: This paper describes the influence of design characteristics on the statistical inference for an ecotoxicological hazard-based model using simulated survival data. The design characteristics of interest are the number and spacing of observations (counts) in time, the number and spacing of exposure concentrations (within cmin and cmax), and the initial number of individuals at time 0 in each concentration. A comparison of the coverage probabilities for confidence limits arising from the profile-likelihood approach and the Wald-based approach is carried out. The Wald-based approach is very sensitive to the choice of design characteristics, whereas the profile-likelihood approach is more robust and unbiased. Special attention is paid to estimating a parametric no-effect concentration in realistic small-sample situations since this is the most interesting parameter from an environmental protection point of view.
TL;DR: In this paper, conditions under which the traditional ANOVA tests are valid are provided, and alternative test procedures are compared to one another in a simulation study, and these alternative procedures include degrees of freedom adjustment techniques, a multivariate model approach, and a mixed models (estimated generalized least squares) approach.
Abstract: Crossover experiments are really special types of repeated measures experiments. The correct analysis of a repeated measures experiment depends on the structure of the variance-covariance matrix of the repeated measures. Certain structures of the variance-covariance matrix assure that analysis of variance procedures lead to valid statistical tests. Other covariance structures invalidate some of the analysis of variance tests. This paper provides conditions under which the traditional ANOVA tests are valid. For those situations when the traditional tests may be invalid, alternative test procedures are proposed, and these procedures are compared to one another in a simulation study. These alternative procedures include degrees of freedom adjustment techniques, a multivariate model approach, and a mixed models (estimated generalized least squares) approach. In cases where the usual ANOVA methods are not valid, the multivariate model approach is the method that met the nominal size requirement for the hypotheses tests of equal treatment and equal carryover effects.
TL;DR: In this article, the authors proposed a McCullagh model for the frequency data from three concentric rings and used this to predict the mean weed density by calibration, and compared their predictions with predictions obtained by using a model based on a homogeneous Poisson process for the weeds.
Abstract: Estimation of the mean weed density in a field can be done by counting the number of weeds in selected survey plots, but since counting is very time consuming, a frequency analysis is often used instead. In a frequency analysis, whether the species is present or not is observed in each survey plot. This method is thus much faster than counting, but you do not obtain an estimate of the mean weed density. Based on data from 25 Danish barley fields, we propose a McCullagh model for the frequency data from three concentric rings and use this to predict the mean weed density by calibration. We discuss alternative models based on spatial point processes for the positions of weeds and compare our predictions with predictions obtained by using a model based on a homogeneous Poisson process for the weeds. The models are evaluated by cross-validation, and it is shown that the proposed model gives predictions with considerably less bias than the Poisson process-based model. Although weed data are considered, the method can, in principle, be used for any botanical frequency analysis of small plants.
TL;DR: In this paper, a model is proposed to analyze data arising from a cross-over experiment in which measurements on a control and an exposed subject are recorded over time within each crossover period.
Abstract: A model is proposed to analyze data arising from a cross-over experiment in which measurements on a control and an exposed subject are recorded over time within each crossover period. The model uses locally weighted quadratic regression to control for nuisance temporal trends common to both control and exposed subjects within each period and specifies a first-order autoregressive process to account for dependence among measurements within each longitudinal sequence. We apply the model to a motivating data set in which laboratory animals are exposed to concentrated air particles.
TL;DR: In this article, a Poisson regression approach is proposed to model small-area variation in source activities and to allocate county or regionwide emission estimates to subcounty units, which is used to model the spatial distribution of automobile refinishing activities in the Sacramento modeling region in California.
Abstract: Urban and regional air quality needs to be analyzed at various geographical scales. Area source emission inventories usually estimate total emissions for various industrial and commercial activities at the county or larger scales. Consequently, information on spatial variation of emissions within a county, a critical requirement for urban airshed modeling, is largely unavailable. This paper proposes a Poisson regression approach that enables us to model small-area variation in source activities and to allocate county or regionwide emission estimates to subcounty units. The new approach is used to model the spatial distribution of automobile refinishing activities in the Sacramento modeling region in California. The paper addresses the problem of overdispersion of variance in Poisson regression and evaluates the effect of grid location on modeling results. The usefulness of geographical information systems in spatial statistical analysis is demonstrated.
TL;DR: The similarities between the four ponds suggest that the results are applicable to similar ponds in Texas and elsewhere throughout the world.
Abstract: We estimate the average amount of phosphorus in the soil of four Texas shrimp ponds. Broad spatial trends in phosphorus levels are observed and the trend is qualitatively common to all four ponds. We calculate and estimate the bias and variance of the estimated mean, accounting for correlations among the measurements. This was done for three different sampling designs and several sampling rates for each design. Using spatial statistical techniques, the diminishing utility of increased sampling is demonstrated (in terms of variance) and the design with the lowest bias is identified. The similarities between the four ponds suggest that the results are applicable to similar ponds in Texas and elsewhere throughout the world.
TL;DR: This paper uses order-restricted inference theory to construct estimators and hypothesis tests for these types of resource selection tests, and uses data from a radio-tracking study on gray partridges to illustrate the techniques.
Abstract: Many studies on animal resource selection involve recording the number of times radio-collared animals are observed in a finite number of resource categories (e.g., habitats). A general objective of these studies is to determine if the animals are using resources disproportionately to resource availability. In this paper, we propose testing ordered resource selections. The advantage of testing ordered resource selections hypotheses is that a researcher can evaluate specific resource selection relationships beyond the multiple comparison testing framework. We use order-restricted inference theory to construct estimators and hypothesis tests for these types of resource selection tests. Detailed illustrations using data from a radio-tracking study on gray partridges illustrate the techniques.
TL;DR: In this article, a linear, heteroscedastic model relating XRF measurements to the lead levels detected using laboratory analysis is presented, which accounts for the fact that the true lead levels were not known by introducing a lognormal measurement error component.
Abstract: X-ray fluorescence (XRF) instruments, used to test for lead in paint, were evaluated in a field study of portable lead measurement technologies, sponsored by the U.S. Environmental Protection Agency and the U.S. Department of Housing and Urban Development. The bias and precision of an XRF instrument were the performance criteria of interest. Estimates of these quantities were obtained using a linear, heteroscedastic model relating XRF measurements to the lead levels detected using laboratory analysis. The model accounts for the fact that the true lead levels were not known by introducing a lognormal measurement error component. This paper discusses properties of the model and its estimation using field study data.
TL;DR: In this article, an extension to a method for estimating population size from recapture data obtained by passive self-marking and concurrent trapping is presented, based on fitting a continuous-time dynamic model of the marking and trapping process, including several special nuisance parameters related to the dynamics of trapping.
Abstract: The self-marking protocol for estimating insect populations has us place both traps (that catch insects permanently) and marking stations (structurally identical to traps but which mark and release); at the end of a trapping period, we recover traps that contain both marked and unmarked individuals. This paper presents extensions to a method for estimating population size from recapture data obtained by passive self-marking and concurrent trapping. The method is based on fitting a continuous-time dynamic model of the marking and trapping process, including several special nuisance parameters related to the dynamics of trapping. The model consists of a series of differential equations, and the numerical solution provides a multinomial distribution for captures in the various marking classes. The model is configured by constraining parameter values that are not to be estimated to their default values, and maximum-likelihood provides an unbiased estimate for most configurations of the model. The notable exception occurs when estimating a differential in the operating rates of marking stations and traps (gl); the procedure tends to overestimate the performance of marking stations and population size. Two-standard-error intervals usually provide the correct coverage except where we estimate gl and the sample is small. Intervals from simple percentiles of a parametric bootstrap improve on coverage.
TL;DR: Both likelihood-based and Bayesian approaches for use in the identification of the site of cleavage of GPI-anchored proteins are developed, demonstrated using four examples where extensive biochemical analyses already have identified the bond in the newly formed polypeptide where a GPI anchor is attached.
Abstract: Many proteins critical to the existence of parasites from infectious agents that affect man, including malaria, leishmania, and trypansoma, are attached to the surface membrane by glycosyl-phosphatidyl inositol (GPI) anchors. Creation of GPI-anchored proteins involves elimination of a bond between adjacent amino acids in a newly formed, or nascent, polypeptide and attachment of a GPI anchor at this site. Because interfering with this process by altering the location of attachment is a potential infection-control strategy in humans, knowledge of the cleavage site in the nascent polypeptide is important. Determination of the cleavage point directly through biochemical analysis is both labor and resource intensive. We develop both likelihood-based and Bayesian approaches for use in the identification of the site of cleavage of GPI-anchored proteins. These methods are demonstrated using four examples where extensive biochemical analyses already have identified the bond in the newly formed polypeptide where a GPI anchor is attached. In contrast to sole reliance on biochemical analyses, use of the proposed methods has great potential for savings of both time and money.
TL;DR: A regression-type classification approach was developed to calculate the cut-off value that could not only successfully separate the two groups of reactions but also significantly reduce the risk of producing false-positive results.
Abstract: The minute virus of mice (MVM) TaqMan assay is an endpoint assay that utilizes the ABI prism 7200 fluorometer system to measure the fluorescence signals (Rn) generated during a reaction called TaqMan assay. In this assay, the virus-free negative samples produce only low-level background signals, i.e., small Rn values, while the virus DNA-containing positive samples generate high signals, i.e., large Rn values. Due to the skewed distribution of Rn values and the fact that the mean Rn values change from plate to plate for both negative and positive samples, it appeared to be impossible to set up the cut-off value that would separate all positive from negative reactions without misclassification using traditional statistical approaches. Using the three-sigma principal and the mean Rn values instead of the individual Rn values, we developed a regression-type classification approach to calculate the cut-off value that could not only successfully separate the two groups of reactions but also significantly reduce the risk of producing false-positive results. It takes into consideration a number of sources of assay variation including intra-assay, interassay, and intercalibration variation; it also considers different losses for false-positive and false-negative results. This approach was tested and proven to be useful for the assay result interpretation in the MVM TaqMan assay validation.
TL;DR: A flexible family of tolerance distributions is fitted to experimental data involving large numbers of insects in a low-oxygen atmosphere and is shown to be clearly superior to the fitting of the probit model.
Abstract: Prediction of the exposure time required for high mortality (99.99-99.999%) is achieved by fitting a flexible family of tolerance distributions to experimental data involving large numbers of insects in a low-oxygen atmosphere. The best fitting member of that family is a double exponential distribution with time untransformed. This statistical model is shown to be clearly superior to the fitting of the probit model. The ability to extrapolate beyond the observed range is discussed.