Experimental Design Issues Research Articles

Microarray Data Analysis: From Hypotheses to Conclusions Using Gene Expression Data

We review several commonly used methods for the design and analysis of microarray data. To begin with, some experimental design issues are addressed. Several approaches for pre‐processing the data (filtering and normalization) before the statistical analysis stage are then discussed. A common first step in this type of analysis is gene selection based on statistical testing. Two approaches, permutation and model‐based methods are explained and we emphasize the need to correct for multiple testing. Moreover, powerful approaches based on gene sets are mentioned. Clustering of either genes or samples is frequently performed when analyzing microarray data. We summarize the basics of both supervised and unsupervised clustering (classification). The latter may be of use for creating diagnostic arrays, for example. Construction of biological networks, such as pathways, is a statistically challenging but complex task that is a relatively new development and hence mentioned only briefly. We finish with some remarks on literature and software. The emphasis in this paper is on the philosophy behind several statistical issues and on a critical interpretation of microarray related analysis methods.

Design considerations for efficient and effective microarray studies.

This article describes the theoretical and practical issues in experimental design for gene expression microarrays. Specifically, this article 1) discusses the basic principles of design (randomization, replication, and blocking) as they pertain to microarrays, and 2) provides some general guidelines for statisticians designing microarray studies.

Using Experimental Auctions for Marketing Applications: A Discussion

The present article discusses general issues associated with experimental auctions and their relative advantages and disadvantages over other marketing research techniques. Experimental auctions create an active market environment with feedback where subjects exchange real goods and real money, which is not generally the case with other methods. The article also discusses four experimental design issues associated with experimental auctions: auction mechanism, market feedback and bidder affiliation, demand reduction and wealth effects, and multiple attribute valuation. Each of these experimental design issues, if not properly controlled, have the potential to create serious flaws in marketing recommendations.

Estimating change rates of genetic markers using serial samples: applications to the transposon IS 6110 in Mycobacterium tuberculosis

In infectious disease epidemiology, it is useful to know how quickly genetic markers of pathogenic agents evolve while inside hosts. We propose a modular framework with which these genotype change rates can be estimated. The estimation scheme requires a model of the underlying process of genetic change, a detection scheme that filters this process into observable quantities, and a monitoring scheme that describes the timing of observations. We study a linear “birth–shift–death” model for change in transposable element genotypes, obtaining maximum-likelihood estimators for various detection and monitoring schemes. The method is applied to serial genotypes of the transposon IS 6110 in Mycobacterium tuberculosis. The estimated birth rate of 0.0161 (events per copy of the transposon per year) and death rate of 0.0108 are both significantly larger than the estimated shift rate of 0.0018. The sum of these estimates, which corresponds to a “half-life” of 2.4 years for a typical strain that has 10 copies of the element, substantially exceeds a previous estimate of 0.0135 total changes per copy per year. We consider experimental design issues that enable the precision of estimates to be improved. We also discuss extensions to other markers and implications for molecular epidemiology.

Microarray expression profiling: capturing a genome-wide portrait of the transcriptome.

The bacterial transcriptome is a dynamic entity that reflects the organism's immediate, ongoing and genome-wide response to its environment. Microarray expression profiling provides a comprehensive portrait of the transcriptional world enabling us to view the organism as a 'system' that is more than the sum of its parts. The vigilance of microorganisms to environmental change, the alacrity of the transcriptional response, the short half-life of bacterial mRNA and the genome-scale nature of the investigation collectively explain the power of this method. These same features pose the most significant experimental design and execution issues which, unless surmounted, predictably generate a distorted image of the transcriptome. Conversely, the expression profile of a properly conceived and conducted microarray experiment can be used for hypothesis testing: disclosure of the metabolic and biosynthetic pathways that underlie adaptation of the organism to chang-ing conditions of growth; the identification of co-ordinately regulated genes; the regulatory circuits and signal transduction systems that mediate the adaptive response; and temporal features of developmental programmes. The study of bacterial pathogenesis by microarray expression profiling poses special challenges and opportunities. Although the technical hurdles are many, obtaining expression profiles of an organism growing in tissue will probably reveal strategies for growth and survival in the host's microenvironment. Identifying these colonization strategies and their cognate expression patterns involves a 'deconstruction' process that combines bioinformatics analysis and in vitro DNA array experimentation.

Diffusion-tensor MRI: theory, experimental design and data analysis - a technical review.

This article treats the theoretical underpinnings of diffusion-tensor magnetic resonance imaging (DT-MRI), as well as experimental design and data analysis issues. We review the mathematical model underlying DT-MRI, discuss the quantitative parameters that are derived from the measured effective diffusion tensor, and describe artifacts that arise in typical DT-MRI acquisitions. We also discuss difficulties in identifying appropriate models to describe water diffusion in heterogeneous tissues, as well as in interpreting experimental data obtained in such issues. Finally, we describe new statistical methods that have been developed to analyse DT-MRI data, and their potential uses in clinical and multi-site studies.

Application of adeno-associated viral vectors in behavioral research

Viral vectors afford the capability of genetically manipulating the expression of neurotransmitters, neuropeptides, hormones, and their receptors in specific brain sites of adult animals of any species. Hence, they are a powerful tool for investigating the neurochemistry underlying complex cognitive processes and behaviors. Here we discuss how the recombinant adeno-associated virus (rAAV) can be engineered for use in neurobehavioral studies, techniques for site-specific delivery of vector into the brain, characterization of expression profiles, and biosafety issues. Finally, we discuss issues of experimental design and interpretation of behavioral results in viral vector studies.

Experimental design issues for the early detection of disease: novel designs.

This paper investigates two experimental designs which have been used to evaluate the benefit of the early detection of breast cancer. They have some advantages over a classical design (the screening program versus usual medical care) in that subjects in a control group may benefit by participating in the study. We refer to the two experimental designs as the up-front (UFD) and close-out (COD) designs. The UFD consists of offering an initial exam to all participants. Then they can be randomized to a usual care group or a screening group receiving one or more special examinations. If the outcome of the initial examination is included in the analysis, then the study can answer the question of the benefit of an additional screening program after an initial examination. If the analysis excludes all the cases diagnosed at the initial examination, then the analysis evaluates the benefit of a screening program after elimination of the prevalent cases. These prevalent cases are most likely to be affected by length bias sampling and consequently will tend to have less aggressive disease and live longer. As a result, the UFD can answer two scientific questions. The COD consists of randomizing subjects to a usual care group and a screened group. However, the usual care group receives an examination which coincides at the time of the last exam in the study group. In this paper the power of these two designs have been evaluated. In both cases the power is severely reduced compared to the usual control group receiving no special exams. The power is a function of the sensitivity of the exam, the number and spacings of the exams given to the screened group as well as the sample size, disease incidence of the population and the survival distribution. The theoretical results on power are applied to the Canadian National Breast Cancer Study (ages 40-49) which used an UFD and the Stockholm Mammography Breast Cancer Screening Trial which utilized a COD.

Gene Expression Analysis Using cDNA Microarrays

AbstractThis unit focuses on cDNA microarrays, which are being increasingly applied to large‐scale gene expression studies in the nervous system, concentrating on the steps of probe preparation and microarray hybridization. Four protocols for the attachment of fluorophores to cDNA (labeling) and cDNA hybridization onto microarrays are presented: direct labeling using either Klenow fragment or reverse transcriptase, and two signal amplification techniques represented by the tyramide signal amplification method (TSA), and detection and labeling of cDNA using PCR amplification. Basic issues in experimental design, sources of variability and potential pitfalls in microarray experiments are described. Anticipated results and follow up of microarray experiments, including confirmation of the results and troubleshooting are also discussed.

Perturbation signal design for neural network based identification of multivariable nonlinear systems

AbstractThe paper focuses on issues in experimental design for identification of nonlinear multivariable systems. Perturbation signal design is analyzed for a hybrid model structure consisting of linear and neural network structures. Input signals, designed to minimize the effects of nonlinearities during the linear model identification for the multivariable case, have been proposed and its properties have been theoretically established. The superiority of the proposed perturbation signal and the hybrid model has been demonstrated through extensive cross validations. The utility of the obtained models for control has also been proved through a case study involving MPC of a nonlinear multivariable neutralization plant.

Optimal Design of Dose Response Experiments: A Model-Oriented Approach

We discuss optimal experimental design issues for nonlinear models arising in dose response studies. The optimization is performed with respect to various criteria which depend on the Fisher information matrix. Special attention is given to models with a variance component that depends on unknown parameters.

FMRI: a new tool for the in vivo localization of drug actions in the brain.

Functional magnetic resonance imaging (fMRI), a still-emerging, non-invasive neuroimaging tool, has been applied to a wide range of questions in sensory, motor control, and cognitive psychology. Only more recently has it been applied to understand the sites and mechanisms of action of pharmacological agents within the human CNS. However, in so doing, a new set of problems and concerns surrounding the technique must be addressed because of the unique transduction mechanisms (both physiological and biophysical) that exist to produce the fMRI signal from the underlying neuronal activity. Experimental design and control issues become paramount in performing fMRI pharmacological protocols and in signal interpretation. With these caveats, the use of pharmacological agents with fMRI is likely to greatly increase in the near term as new questions about both brain physiology and neuropharmacological mechanisms become addressable for the first time. Examples are given using nicotine and cocaine as a prototypical agents.

Experimental design for gene expression microarrays.

We examine experimental design issues arising with gene expression microarray technology. Microarray experiments have multiple sources of variation, and experimental plans should ensure that effects of interest are not confounded with ancillary effects. A commonly used design is shown to violate this principle and to be generally inefficient. We explore the connection between microarray designs and classical block design and use a family of ANOVA models as a guide to choosing a design. We combine principles of good design and A-optimality to give a general set of recommendations for design with microarrays. These recommendations are illustrated in detail for one kind of experimental objective, where we also give the results of a computer search for good designs.

Functional magnetic resonance imaging

Functional magnetic resonance imaging (fMRI) is a relatively new technique for imaging changes in brain haemodynamics linked to local neuronal activity. It does not entail exposure to radioactivity and has other important technical advantages over previously available methods for mapping functional neuroanatomy, such as positron emission tomography. It is likely that fMRI will become an increasingly important and widely used technique for psychiatric research in the next 5 years; applications to clinical practice are also conceivable. Here we review the technical foundations of fMRI. We begin with a brief summary of the basic physical principles of magnetization and nuclear magnetic resonance. This is followed by a description of some commonly used procedures for magnetic resonance imaging in general; an introduction to exogenous and endogenous contrast agents; and a section on gradient echo echoplanar imaging for fMRI specifically. We describe common artefacts and essential hardware requirements. Then issues of experimental design and data analysis are discussed in conclusion. Given the rapid rate of scientific progress in this field, this review can be no more than introductory. References to some supplementary textbooks and specialist journals are provided as guides to further reading.

Comprehensive Versus Matched Psychosocial Treatment in the MTA Study: Conceptual and Empirical Issues

Addressed some factual inaccuracies and presented alternative positions on key issues raised in the article by Greene and Ablon (this issue) on the question, "What does the Multimodal Treatment Study (MTA) tell us about effective psychosocial treatment for attention deficit hyperactivity disorder (ADHD)?" The Greene and Ablon critique does not present for the reader's consideration the full range of findings from the MTA study, notably those most relevant to psychosocial treatment, and articulates a theoretical position that effective treatment requires matching treatment to children's assessed needs, an approach not taken in the MTA study. In this article, I present the full range of findings from the MTA study related to psychosocial treatment effects, correct the misperceptions that exist about the study based on limited reviews such as Greene and Ablon's, and review the empirical and experimental design issues that produced the decision by the MTA investigative team to study the effects of intensive, comprehensive psychosocial treatment. I argue that the questions asked by the MTA study about psychosocial treatment were important, relevant, and were addressed well in the MTA study design.

Statistical design and the analysis of gene expression microarray data.

Gene expression microarrays are an innovative technology with enormous promise to help geneticists explore and understand the genome. Although the potential of this technology has been clearly demonstrated, many important and interesting statistical questions persist. We relate certain features of microarrays to other kinds of experimental data and argue that classical statistical techniques are appropriate and useful. We advocate greater attention to experimental design issues and a more prominent role for the ideas of statistical inference in microarray studies.

Functional magnetic resonance imaging

Functional magnetic resonance imaging (fMRI) is a relatively new technique for imaging changes in brain haemodynamics linked to local neuronal activity. It does not entail exposure to radioactivity and has other important technical advantages over previously available methods for mapping functional neuroanatomy, such as positron emission tomography. It is likely that fMRI will become an increasingly important and widely used technique for psychiatric research in the next 5 years; applications to clinical practice are also conceivable. Here we review the technical foundations of fMRI. We begin with a brief summary of the basic physical principles of magnetization and nuclear magnetic resonance. This is followed by a description of some commonly used procedures for magnetic resonance imaging in general; an introduction to exogenous and endogenous contrast agents; and a section on gradient echo echoplanar imaging for fMRI specifically. We describe common artefacts and essential hardware requirements. Then issues of experimental design and data analysis are discussed in conclusion. Given the rapid rate of scientific progress in this field, this review can be no more than introductory. References to some supplementary textbooks and specialist journals are provided as guides to further reading.

Intercomparison of Arctic regional climate simulations: Case studies of January and June 1990

Advances in regional climate modeling must be strongly based on analysis of physical processes in comparison with data. In a data‐poor region such as the Arctic; this procedure may be enhanced by a community‐based approach, i.e., through collaborative analysis by several research groups. To illustrate this approach, simulations were performed with two regional climate models, HIRHAM and ARCSyM, over the Arctic basin to 65°N, laterally driven at the boundaries by observational analyses. It was found that both models are able to reproduce reasonably the main features of the large‐scale flow and the surface parameters in the Arctic. Distinct differences in the simulations can be attributed to specific characteristics of the boundary layer and surface parameterizations, which result in surface flux differences, and to the lateral moisture forcing, both of which affect moisture availability in the atmosphere. Further disparities are associated with the additional degrees of freedom allowed in the coupled model ARCSyM. Issues of model configuration and experimental design are discussed, including domain size, grid spacing, boundary formulations, model initialization and spin‐up, and ensemble approaches. In order to reach definitive conclusions in a regional climate model intercomparison, ensemble simulations with adequate spin‐up and equivalent initialization of surface fields will be required.

Induction of Proliferative Lesions of the Uterus, Testes, and Liver in Swiss Mice Given Repeated Injections of Sodium Arsenate: Possible Estrogenic Mode of Action

Inorganic arsenic (As) is a human carcinogen but has not been unequivocally proven carcinogenic in rodents. For instance, one older study indicates that repeated iv injections of sodium arsenate might induce lymphomas in Swiss mice (58% incidence) (Osswald and Goerttler, Verh. Dtsch. Ges. Pathol. 55, 289–293, 1971), but it was considered inadequate for critical evaluation of carcinogenic potential largely because of issues in experimental design. Therefore, we studied repeated iv sodium arsenate injection and neoplastic response in male and female Swiss mice. Groups (n = 25) of mice received sodium arsenate (0.5 mg/kg, iv) or saline (control) once/week for 20 weeks and were observed for a total of 96 weeks when the study ended. Differences in survival and body weights were unremarkable. In females, arsenate induced marked increases in the incidence and severity of cystic hyperplasia of the uterus compared against controls. Arsenate also was associated with a rare adenocarcinoma of the uterus. Hyperplastic uterine epithelium from arsenate-exposed animals showed strong positive immunostaining for the proliferating cell nuclear antigen (PCNA). There was also an upregulation of estrogen receptor (ER) immunoreactive protein in the early lesions of uterine luminal and glandular hyperplasia, although a progressive decrease in its expression was seen in the severe hyperplastic or neoplastic epithelium. In common with the preneoplastic and neoplastic gynecological lesions in humans, the levels of immunoreactive inducible nitric oxide synthase (iNOS) and 3-nitrotyrosine-containing proteins were greater in the uterine hyperplastic epidermis and their intensity was positively correlated with the severity of the lesions. Arsenate-induced uterine hyperplastic lesions also showed a strong upregulation of cyclin D1, an estrogen-associated gene product essential for progression through the G1 phase of the cell cycle. In other tissues, arsenate increased testicular interstitial cell hyperplasia incidence and severity over control but without affecting the incidence of tubular degeneration. Arsenate also induced increases in hepatic proliferative lesions (HPL; foci of alteration + neoplasia), but only in females. Significant skin changes (incidence of hyperkeratotic lesions) and renal lesions (severity of nephropathy) also occurred in arsenate-treated females. Thus, repeated arsenate exposure, though not outright tumorigenic in the present study, was associated with proliferative, preneoplastic lesions of the uterus, testes, and liver. Estrogen treatment has been associated with proliferative lesions and tumors of the uterus, female liver, and testes in other studies, supporting a hypothesis that arsenate might somehow act through an estrogenic mode of action.

Measurement of Disposition Half‐Life, Clearance, and Residence Times

This unit provides an overview of the principles underlying the elimination process. Experimental design issues, methods of data analysis, and complications that may be encountered are discussed. A complete worked example is provided to exemplify the methods of analysis that are available.