Appendixes | Glossary

A numerical identifier for a Gene, marker, or gene product in a public database. CDISC Analysis Data Model. CDISC ADaM Subject-Level Analysis Data set. Any adverse change in health or side effect that occurs in a person who participates in a Clinical Research trial while the patient is receiving treatment or within a previously specified length of time after treatment completion. A transaminase enzyme, also known as serum glutamic pyruvic transaminase (SGPT) or alanine aminotransferase (ALAT). Catalyzing two parts of the alanine cycle, it is found in body tissues (most commonly associated with liver tissues) and serum. It is a clinical diagnostic indicator of hepatocellular injury. Significance level. Although alpha can be any value between 0 and 1, it is typically set at either 0.01, 0.05 or 0.10. A position that a researcher evaluates in an experiment. The alternative hypothesis, H1 (or Ha), is the hypothesis that sample observations are influenced by a specific non-random cause. It is rival to the Null Hypothesis, H0. The building block of Proteins. An enormous variety of proteins can be constructed by linking together Amino Acids (there are more than twenty types) in a linear chain, which in turn become folded in more complex configurations. Analysis of covariance; a general linear model with a continuous outcome Variable and multiple predictor variables, with at least one nominal and one continuous predictor variable. Considered a hybrid of regression for continuous variables and ANOVA, ANCOVA can determine whether specific factors have an impact on the outcome variable after removing variance resulting from Covariates (the qualitative predictors). Statistical models and procedures that partition observed Variance in a Variable into components attributable to different variation sources. By analyzing comparisons of variance estimates, ANOVA can determine whether the Means of several groups are statistically equal. In a Clinical Research trial, the group of patients receiving a certain type of therapy. For example, one arm of a clinical trial might consist of patients receiving a new medication, another arm might consist of a standard-of-care medication, and another a placebo pill. A pyridoxal phosphate-dependent transaminase enzyme, also known as aspartate aminotransferase (AspAT, ASAT, AAT) or serum glutamic oxaloacetic transaminase (SGOT). Catalyzing a reversible transfer between aspartate and glutamate, it is found in the brain, heart, muscles, kidneys, red blood cells, and liver. It is a clinical diagnostic indicator of liver health. A group of related of functionally similar Observations that are considered as a unit for statistical analysis. A regression method where the Dependent Variable contains binomial values (for example, 0 and 1, often corresponding to ‘no’ and ‘yes’, or ‘failure’ and ‘success’, respectively). A group of Organs that work together to perform a task. Examples in humans include the digestive system, the nervous system, and the endocrine system. The practice of using with-replacement empirical distributions of Observations to estimate the statistical properties of the population from which the observations were made. Used to display the response distribution at different combinations of factor levels. Box plots can reveal differences in the response Mean at different levels, suggesting Main Effects. Box plots can also reveal whether the response variation is homogenous across factor levels, an assumption made in ANOVA. See Box Plot. A two-dimensional Scatterplot showing the relationship between two Variables over time. Each circle, or bubble, represents a single instance of an ID variable. See Bubble Plot. An optional (in most APs) Variable specification whose values define groups of Observations, such as hour, month, or year. Specifying a BY variable enables you to animate an image so that you can see how response values change according to some grouping, like over time. Alternatively, BY variables can enable analyses to be performed separately on different groups as defined by a variable such as gender. Clinical Data Interchange Standards Consortium, a nonprofit organization that has “established standards to support the acquisition, exchange, submission, and archive of Clinical Research data and Metadata” whose mission is “to develop and support global, platform-independent data standards that enable information system interoperability to improve medical research and related areas of health-care”. See the CDISC website for more information. These columns specify those Observations for which data have been censored or truncated. For example, investigations of the effects of certain Genes on life span might be terminated before all of the individuals have expired. The ultimate life spans for these individuals are unknown. All that can be said is that they exceed the period of the study. These data are considered censored. A statistical test used to test the existence of a relationship between two nominal Variables where the sampling distribution of the Test Statistic is a chi-squared distribution when the Null Hypothesis is true (or where it is asymptotically true). Bundled strands of DNA and Protein located in the Cell nucleus. Chromosomes are inherited from parents. Chromosome count per organism cell varies by species. Human cells contain 23 pairs of nuclear chromosomes. The process of dividing a data set into mutually exclusive groups such that the Observations for each group are as close as possible to one another, and different groups are as far as possible from one another. The probability of an event (for example, X) given that another specific event (for example, Y) occurs. Conditional probability is often expressed as P(X|Y) or PY(X). A relationship between Variables in terms of dependence. Also known as the Pearson product-moment correlation coefficient, it is equal to the Covariance of two Variables divided by the product of their Standard Deviations. A measure of the relationship between two Variables. It equals the Correlation Coefficient between the two variables times the square roots of their Variances. An Independent Variable, not manipulated by the experimenter, that can influence the outcome of the experiment. A classical semiparametric (sometimes considered nonparametric) method that relates the time of an event (for example, failure or death) to explanatory variables (Covariates). This model assumes that hazard rate, rather than survival time, is a function of the explanatory variables. There are no assumptions made on the shape or nature of the hazard function. Comma-separated value format. This text format stores tabular data, with line breaks and commas used to delimit table rows and columns, respectively. One or more characters that separate the designations for the different Alleles in a genotype. JMP frequently uses a forward-slash (/) as a delimiter. A tree-like diagram used to summarize a Clustering report. A dendrogram shows where each cluster divides in a hierarchical fashion. See Dendrogram. Deoxyribonucleic acid, a nucleic acid containing genetic instructions for the development and functioning of living organisms (except for RNA viruses). DNA segments encoding genetic information are known as Genes. Non-coding DNA can have structural or regulatory purposes. An algebraic operation that takes two equal-length number sequences (usually coordinate vectors) and returns a single number obtained by multiplying corresponding entries and summing those products. Double False Discovery Rate (FDR) Adjustment The Double FDR method of Mehrotra and Heyse (2004)^{2}is used to compare the incidence of adverse events among treatments, leveraging the grouping of related adverse events (typically defined by the MedDRA system organ class). The method considers whether related terms within a group show differences between the treatments and upweights or downweights the significance of an individual term within the group accordingly. In the 2004 paper, the FDR adjustment is performed twice, and simulations are used to control the false discovery rate. Mehrotra and Adewale (2011) refine the Double FDR method to avoid the need for simulations by applying FDR adjustment thrice. For a given square matrix, a nonzero vector that changes length, but not direction, when multiplied by the matrix. The computation of principal components for a set of Variables uses the eigenvectors of the variables' Covariance or Correlation matrix. Also referred to as an Independent Variable or predictor variable, a factor is a Variable included in a model to account for variation in a response. Factors are the variables whose values (levels) you set to study their relationship to a response. You often experiment with many potentially influential factors at the same time. The probability of making one or more false discoveries (Type I Errors) among all hypotheses while performing multiple pairwise tests. A statistical significance test used in the analysis of contingency tables where sample sizes are small. It is useful when you want to conduct a Chi-square Test, but one of your cells has an expected frequency of five or less. Its name is derived from its inventor, R.A. Fisher, and reflects that the significance of the deviation from a Null Hypothesis can be calculated exactly (as opposed to relying on an approximation whose exactness is realized only as sample size approaches infinity). A graphical display designed to illustrate the relative strength of treatment effects (or relative degree of gene enrichment), in multiple quantitative scientific studies (or databases) addressing the same question. Forest plots generally display results for each study (or other data source) as horizontal lines representing the 95% confidence interval of the effect observed in that trial. See Forest Plot. Molecular unit of heredity in a living organism, comprising a contiguous sequence of DNA or RNA, coding for a Protein or RNA chain, which in turn has a function in the organism. A method of cluster analysis that constructs a hierarchy of clustering. Strategies include the agglomerative approach, where each cluster initially contains only one observation, and the divisive approach, where all observations are initially contained in one cluster. Hierarchical clustering results are commonly presented in Dendrogram form. The second-highest level of the Medical Dictionary for Regulatory Activities (MedDRA) Hierarchy, below System Organ Class (SOC) and above High Level Term (HLT). An example of an HLGT is “Respiratory tract infections”. The third-highest level of the Medical Dictionary for Regulatory Activities (MedDRA) Hierarchy, below High Level Group Term (HLGT) and above Preferred Term (PT). An example of an HLT is “Viral upper respiratory tract infections”. A nonparametric measure of association that detects general departures from independence. This Statistic approximates a weighted sum over observations of chi-square statistics for two-by-two classification tables. A test of the Null Hypothesis that “the population Mean vector is equal to the given mean vector”. It is the multidimensional equivalent of the one-sample t-test. A decision-making rule based on data from an experiment or observational study. A hypothesis test is used to conclude significance of a result based on the sufficiently low likelihood (set by the predefined significance level) that it occurred because of random chance alone. An ominous prognostic indicator (in Clinical Research) that a pure drug-induced liver injury (DILI) leading to jaundice, without a hepatic transplant, has a case fatality rate of 10-50%. One or more columns specifying how the Observations are to be classified. In JMP, a journal is a file (.jrn) that contains results of user-specified reports. A curve based on the survival function estimator from life-time or clinical outcome data. For example, it can be used to measure the proportion of patients living for a given amount of time after treatment, or to measure the time until a tumor disappears. A statistical method that creates optimally separated groups of Observations in data using one of several methods. A set of points called cluster seeds is selected as a first guess of the means of the clusters. One cluster seed is selected for each of k clusters. Each observation is assigned to the nearest seed to form temporary clusters. The seeds are then replaced by the Means of the temporary clusters, and the report is repeated until no further changes occur in the clusters. A column containing descriptive labels that can be printed in the output by certain procedures instead of, or in addition to, the Variable name (which is also known as the SAS Variable Name). Synonymous with SAS Variable Label. A successive hierarchical partition of data in a Tree Map. The first level represents the entire unpartitioned data set. The second level represents the first partition of the data into segments, and so on. A nonparametric Hypothesis Test to compare the survival distributions of two samples. It is appropriate when data are right-skewed and non-informatively censored. Also known as a Mantel-Cox test. This test can be considered a time-stratified Cochran-Mantel-Haenszel Test. A generalized linear model used for prediction of the probability of event occurrence (Binomial Regression) by fitting data to a Logit Function logistic curve. The lowest level of the Medical Dictionary for Regulatory Activities (MedDRA), below Preferred Term (PT). This level is reserved for non-current, vague, ambiguous, truncated, or misspelled terms, or for terms taken from other terminologies that do not conform to MedDRA rules. Least squares Means, which are estimates of means of classification effects that would be observed, assuming that the experimental design is balanced. A distance measure based on Correlations between Variables. In contrast to Euclidean Distance, it is better adapted to non-spherically symmetric distributions, and is scale-invariant. See Mahalanobis Distances. An effect measures the extent to which the response depends on the factors involved in the effect. A main effect is the change in the response due to a single factor. For two-level factors, the main effect is the difference between the mean response at the high level of a factor and the Mean response at its low level. Multivariate analysis of covariance; an extension of the analysis of covariance (ANCOVA) for multiple Dependent Variables or where it is not feasible to combine dependent variables. MANCOVA is similar to MANOVA, but enables control for additional continuous Independent Variables (Covariates). Multivariate analysis of variance, a generalized form of univariate analysis of variance (ANOVA), used when there are two or more Dependent Variables. This analysis is useful in determining whether changes in the Independent Variables have significant effects on the dependent variables, as well as the associated interactions among dependent and independent variables. A Clinical Research comparison of average score during baseline and a summary score during the trial for each finding. See Matched Pairs Analysis. Mathematical average for a collection of n Observations. It is calculated by dividing the sum of the observations by n. In any set of n Observations arranged in order of magnitude, the median is represented by the observation positioned at n/2. The primary list of items at the top of a , which represent the actions or classes of actions that can be executed. Selecting an item executes an action, opens a Pull-down Menu, or opens a Dialog box that requests additional information. A value in the SAS System indicating that no data is stored for the Variable in the current Observation. It is indicated by a single dot (.) for a numeric variable or a blank for a character variable. An argument of proof by contradiction; often known as denying the consequent. It has the general argument form of:2. Not Q. A graphical representation of a two-way frequency table or Contingency Table. A mosaic plot is divided into colored rectangles, so that the area of each rectangle is proportional to the proportions of the Y Variable in each level of the X Variable. See Mosaic Plot. A general or default position that a researcher tests (and attempts to reject) in an experiment. The null hypothesis, H0, is an essential part of a research design, and usually proposes that sample Observations result purely from chance. The null hypothesis can never be proven; data can reject it or fail to reject it only. If a null hypothesis is rejected, an Alternative Hypothesis, H1 (or Ha) is accepted. Analysis of variance with one between-groups factor. This is useful when you have a nominal Independent Variable and a normally distributed interval Dependent Variable, and you want to compare differences in the means of the dependent variable according to levels of the independent variable. A plot showing the response points along the Y axis for each X factor value. Using the plot, you can compare the distribution of the response across the levels of the X factor. The distinct values of X are sometimes called levels. See One-way Plot. Analysis of variance used on one nominal Independent Variable and a normally distributed interval Dependent Variable that is repeated at least twice for each subject. This method is equivalent to the paired samples t-test, but allows for two or more nominal variable levels. A collection of Tissues combined in one structure to serve a particular function. A plot showing several lines or markers on the Y axis overlaid to a common variable on the X axis. See Overlay Plot. A parametric measure of association for two variables. It measures both the strength and the direction of a linear relationship. If one variable X is an exact linear function of another variable Y, a positive relationship exists if the correlation is 1 and a negative relationship exists if the correlation is -1. If there is no linear predictability between the two variables, the correlation is 0. If the two variables are normal with a correlation 0, the two variables are independent. However, correlation does not imply causality because, in some cases, an underlying causal relationship might not exist. A discriminative classifier known for simultaneous Variable selection and classification. Its performance declines as the number of variables increases, and is often compared with that of Support Vector Machine (SVM). The value of a Variable below which a certain percent of Observations fall. For example, the 60th percentile is the value below which 60% of the observations can be found. Note the following percentile landmarks. The Conditional Probability of a random event that is assigned after relevant evidence is considered. Contrast with Prior Probability. The probability of a statistical significance test enabling you to reject the Null Hypothesis when the Alternative Hypothesis is true. Power equals one minus Beta (the rate of Type II Error). The fourth-highest level of the Medical Dictionary for Regulatory Activities (MedDRA) Hierarchy, below High Level Term (HLT) and above Lowest Level Term (LLT). An example of a PT is “Influenza”. The list of menu items or choices that appears when you choose an item from a Menu Bar or from another menu. The statistical probability that a Statistic is as or more extreme than the observed value, assuming that the Null Hypothesis is true. A smaller p-value enables you to more rigorously reject the null hypothesis. The n x p population structure incidence matrix where n is the number of individuals assayed and p is the number of populations defined. Techniques for modeling and analyzing several Variables, with the focus on the relationship between dependent and independent variables. Regression analysis is useful in uncovering how values of a Dependent Variable change when a single Independent Variable is varied. A graph where the conditional distribution of the observations, given the forecast probability, is plotted against the forecast probability. The distributions for perfectly reliable forecasts are plotted along the 45-degree diagonal. See Reliability Diagram. The number of Observations that constitute a statistical sample. For example, the sample size in a study might consist of the number of subjects. Greater sample sizes lead to greater precision and Power for a study design to detect an effect at a given size. Variables (columns) in a SAS data set can have a SAS Variable Label. This label has much less restrictive creation rules than the corresponding SAS Variable Name. Blank spaces, special characters, and longer lengths are permitted. A graph showing the relationship between two Variables. Multiple scatterplot formats exist, including scatterplot matrices, three-dimensional scatterplots, and Bubble Plots. See Reliability Diagram. A subject in a Clinical Research study that skips treatments or otherwise does not meet treatment criteria. In clinical data sets, a value of “Screen Failure” is given in the treatment column for this subject. The factorization of a real or complex matrix, allowing the matrix to be expressed as a product. Every m x n matrix has a singular value decomposition. Nonparametric method for examining whether two quantitative Variables co-vary. Each pair of variables is converted to ranks and is linked with an “unseen nominal variable. The standard deviation of the sample mean. It is calculated by dividing the Standard Deviation by the square root of the Sample Size. A group of Preferred Term (PT)s and Lowest Level Term (LLT)s relating to a particular medical condition or concept. It could also include High Level Term (HLT)s and High Level Group Term (HLGT)s, as well as hierarchies. Such a grouping is helpful in formulating a “case definition” and in data exploration, search, and retrieval. Each medical condition or group of related conditions has one or more individual SMQs. Terms listed in the SMQ define signs, symptoms, events, laboratory data, physical and physiological findings, and so on. The highest level of the Medical Dictionary for Regulatory Activities (MedDRA) Hierarchy, above High Level Group Term (HLGT). An example of an SOC is “Respiratory, thoracic, and mediastinal disorders”. A SAS data set that has samples as columns and molecular entity (for example, marker, gene, clone, protein, or metabolite) as rows. Tall data sets are the transpose of Wide Data Sets. A nonparametric measure of association based on the number of concordances and discordances in paired Observations. Concordance occurs when paired observations vary together, and discordance occurs when paired observations vary differently. Also, used for the truncated product p-Value adjustment method to indicate that there is at least one false Null Hypothesis among those with p-values less than tau when the null hypothesis is rejected. A function of the data sample that reduces and summarizes the data to either one or a few values that can be used to conduct a Hypothesis Test. A collection of Cells from the same origin that achieves a specific function. The cells do not need to be identical in morphology or genomic content, but must have identical function. Examples of animal and plant tissues include muscle and meristematic tissues, respectively. The process of applying a function to a Variable in order to adjust the variable's range, variability, or both. A method that smooths p-Values over s of markers for n Hypothesis Tests by taking the product of those p-values less than a specified cutoff value and evaluating the probability of this product under the overall hypothesis that all n hypotheses are true. A test that assesses the statistical difference between the Means of two different experimental groups. The Test Statistic follows a Student’s t distribution if the Null Hypothesis is supported.If only one Variable is chosen (one-sample t-test), the null hypothesis is that “the population mean is equal to the given mean”. An incorrect decision made when a test rejects a true Null Hypothesis (H0). This is comparable to a false positive error. Type I error rate is denoted by Alpha, and is referred to as the size of the test. An incorrect decision made when a test fails to reject a false Null Hypothesis (H0). This is comparable to a false negative error. Type II error rate is denoted by Beta, and is related to the Power of a test (power = 1-beta). A Scatterplot of the negative log10-transformed p-Values derived from a specific t-test against the log2-fold change in expression. Genes whose expression is decreased lie to the left of the Mean; genes whose expression is increased lie to the right of the mean. Genes with statistically significant differential expression lie above a horizontal threshold. This plot provides an effective means for visualizing the direction, magnitude, and significance of changes in gene expression. See Volcano Plot. A SAS statement that enables you to filter a set of Observations so that only the subset of data meeting the specific filtering criteria are considered in the analysis. A SAS data set that has samples as rows and molecular entity (for example, marker, gene, clone, protein, or metabolite) as columns. Wide data sets are the transpose of Tall Data Sets. A nonparametric statistical hypothesis test used when comparing two related samples or repeated measurements on a single sample to assess whether their population Mean ranks differ. It is appropriate as an alternative to the paired Student’s t-test when the population is not normally distributed or the data is ordinal. A series of processes run in a specified order, whose output is collected in a Journal. Given a constant basic experimental design and analysis objectives, a workflow can be used repeatedly with different data sets.

Bate, A, et al. (1998) A Bayesian neural network method for adverse drug reaction signal generation. Eur J Clin Pharmacol 54:315-321; Gould, AL. (2003) Practical pharmacovigilance analysis strategies. Pharmacoepidemiology and Drug Safety 12: 559–574.

Mehrotra DV, Heyse JH. (2004) Use of the false discovery rate for evaluating clinical safety data. Statistical Methods in Medical Research 13:227-238.

DuMouchel W. (1999) Bayesian data mining in large frequency tables with an application to the FDA spontaneous reporting system. The American Statistician 53: 177-90.

Evans SJW, Waller PC, Davis S. (2001) Use of proportional reporting ratios (PRRs) for signal generation from spontaneous adverse drug reaction reports. Pharmacoepidemiology and Drug Safety 10:483-486.

Meyboom RHB, Egberts ACG, Edwards IR, Hekster YA, de Koning FHP, Gribnau FWJ. (1997) Principles of signal detection in pharmacovigilance. Drug Safe 16: 355–365.