Interaction Biases in Two-Period Crossover Studies: A Modified Analysis to Test with More Sensitivity

The two-period crossover trial is considered the most powerful means of determining the efficacy of new drugs. However, this study design is frequently invalidated by treatment-by-period interaction. If, for example, the effect of the first treatment period carries on into the next one, then it influences the response to the latter period (carryover effect). A second problem is, that the standard approach (Hills-Armitage analysis) for interaction bias has a low statistical sensitivity. The author recently described an alternative method entitled the clinical approach because it looks at the clinical performance of the separate treatment groups and not, as the standard approach, at the means of the groups. It may be hypothesized that this alternative approach is statistically more sensitive than the standard in situations where there is interaction in just one of the treatment groups. The present study uses two examples and a mathematical model. It shows that in case of single-group interaction the clinical approach can, indeed, detect carryover effect at a 30% lower level than the standard. On the other hand, however, the standard approach does so even at a 40% lower level than the clinical in case of two-group interaction. I conclude that one approach supplements the other and that they be used in future studies simultaneously.     

Direct estimation of genetic principal components: simplified analysis of complex phenotypes

Estimating the genetic and environmental variances for multivariate and function-valued phenotypes poses problems for estimation and interpretation. Even when the phenotype of interest has a large number of dimensions, most variation is typically associated with a small number of principal components (eigen-vectors or eigenfunctions). We propose an approach that directly estimates these leading principal components; these then give estimates for the covariance matrices (or functions). Direct estimation of the principal components reduces the number of parameters to be estimated, uses the data efficiently, and provides the basis for new estimation algorithms. We develop these concepts for both multivariate and function-valued phenotypes and illustrate their application in the restricted maximum-likelihood framework.     

On the Analysis of Genome-Wide Association Studies in Family-Based Designs: A Universal,Robust Analysis Approach and an Application to Four Genome-Wide Association Studies

For genome-wide association studies in family-based designs, we propose a new, universally applicable approach. The new test statistic exploits all available information about the association, while, by virtue of its design, it maintains the same robustness against population admixture as traditional family-based approaches that are based exclusively on the within-family information. The approach is suitable for the analysis of almost any trait type, e.g. binary, continuous, time-to-onset, multivariate, etc., and combinations of those. We use simulation studies to verify all theoretically derived properties of the approach, estimate its power, and compare it with other standard approaches. We illustrate the practical implications of the new analysis method by an application to a lung-function phenotype, forced expiratory volume in one second (FEV1) in 4 genome-wide association studies.     

TATES: Efficient Multivariate Genotype-Phenotype Analysis for Genome-Wide Association Studies

To date, the genome-wide association study (GWAS) is the primary tool to identify genetic variants that cause phenotypic variation. As GWAS analyses are generally univariate in nature, multivariate phenotypic information is usually reduced to a single composite score. This practice often results in loss of statistical power to detect causal variants. Multivariate genotype–phenotype methods do exist but attain maximal power only in special circumstances. Here, we present a new multivariate method that we refer to as TATES (Trait-based Association Test that uses Extended Simes procedure), inspired by the GATES procedure proposed by Li et al (2011). For each component of a multivariate trait, TATES combines p-values obtained in standard univariate GWAS to acquire one trait-based p-value, while correcting for correlations between components. Extensive simulations, probing a wide variety of genotype–phenotype models, show that TATES''s false positive rate is correct, and that TATES''s statistical power to detect causal variants explaining 0.5% of the variance can be 2.5–9 times higher than the power of univariate tests based on composite scores and 1.5–2 times higher than the power of the standard MANOVA. Unlike other multivariate methods, TATES detects both genetic variants that are common to multiple phenotypes and genetic variants that are specific to a single phenotype, i.e. TATES provides a more complete view of the genetic architecture of complex traits. As the actual causal genotype–phenotype model is usually unknown and probably phenotypically and genetically complex, TATES, available as an open source program, constitutes a powerful new multivariate strategy that allows researchers to identify novel causal variants, while the complexity of traits is no longer a limiting factor.     

MultiPhen: joint model of multiple phenotypes can increase discovery in GWAS

The genome-wide association study (GWAS) approach has discovered hundreds of genetic variants associated with diseases and quantitative traits. However, despite clinical overlap and statistical correlation between many phenotypes, GWAS are generally performed one-phenotype-at-a-time. Here we compare the performance of modelling multiple phenotypes jointly with that of the standard univariate approach. We introduce a new method and software, MultiPhen, that models multiple phenotypes simultaneously in a fast and interpretable way. By performing ordinal regression, MultiPhen tests the linear combination of phenotypes most associated with the genotypes at each SNP, and thus potentially captures effects hidden to single phenotype GWAS. We demonstrate via simulation that this approach provides a dramatic increase in power in many scenarios. There is a boost in power for variants that affect multiple phenotypes and for those that affect only one phenotype. While other multivariate methods have similar power gains, we describe several benefits of MultiPhen over these. In particular, we demonstrate that other multivariate methods that assume the genotypes are normally distributed, such as canonical correlation analysis (CCA) and MANOVA, can have highly inflated type-1 error rates when testing case-control or non-normal continuous phenotypes, while MultiPhen produces no such inflation. To test the performance of MultiPhen on real data we applied it to lipid traits in the Northern Finland Birth Cohort 1966 (NFBC1966). In these data MultiPhen discovers 21% more independent SNPs with known associations than the standard univariate GWAS approach, while applying MultiPhen in addition to the standard approach provides 37% increased discovery. The most associated linear combinations of the lipids estimated by MultiPhen at the leading SNPs accurately reflect the Friedewald Formula, suggesting that MultiPhen could be used to refine the definition of existing phenotypes or uncover novel heritable phenotypes.     

Rethinking historical controls

Inference from traditional historical controls, i.e. comparing a new treatment in a current series of patients with an old treatment in a previous series of patients, may be subject to a strong selection bias. To avoid this bias, Baker and Lindeman (1994) proposed the paired availability design. By applying this methodology to estimate the effect of epidural analgesia on the probability of Cesarean section, we made two important contributions with the current study. First, we generalized the methodology to include different types of availability and multiple time periods. Second, we investigated how well the paired availability design reduced selection bias by comparing results to those from a meta-analysis of randomized trials and a multivariate analysis of concurrent controls. The confidence interval from the paired availability approach differed considerably from that of the multivariate analysis of concurrent controls but was similar to that from the meta-analysis of randomized trials. Because we believe the multivariate analysis of concurrent controls omitted an important predictor and the meta-analysis of randomized trials was the gold standard for inference, we concluded that the paired availability design did, in fact, reduce selection bias.     

A Multivariate Granger Causality Concept towards Full Brain Functional Connectivity

Detecting changes of spatially high-resolution functional connectivity patterns in the brain is crucial for improving the fundamental understanding of brain function in both health and disease, yet still poses one of the biggest challenges in computational neuroscience. Currently, classical multivariate Granger Causality analyses of directed interactions between single process components in coupled systems are commonly restricted to spatially low- dimensional data, which requires a pre-selection or aggregation of time series as a preprocessing step. In this paper we propose a new fully multivariate Granger Causality approach with embedded dimension reduction that makes it possible to obtain a representation of functional connectivity for spatially high-dimensional data. The resulting functional connectivity networks may consist of several thousand vertices and thus contain more detailed information compared to connectivity networks obtained from approaches based on particular regions of interest. Our large scale Granger Causality approach is applied to synthetic and resting state fMRI data with a focus on how well network community structure, which represents a functional segmentation of the network, is preserved. It is demonstrated that a number of different community detection algorithms, which utilize a variety of algorithmic strategies and exploit topological features differently, reveal meaningful information on the underlying network module structure.     

Using size distributions to detect nutrient and sediment effects within and between habitats

We studied the use of size distributions as a response variable in limnological experiments. Previous quantifications of size distributions were incomplete or difficult to use in experimental settings, and we developed a multivariate approach that more fully describes the shape and biomass of planktonic and benthic size distributions. We re-evaluate the hypothesis that fish affect the shape and nutrients affect the biomass of size distributions, and show that the multivariate approach is more responsive to detecting treatment effects. In a mesocosm experiment, we use this new quantification and analysis of size distributions to detect the main and interactive effects of nutrient addition and sediment type on both benthic and pelagic size distributions. Size distributions in both habitats responded to the nutrient and sediment treatments, indicating linkage since a treatment applied in one habitat affected the size distribution in the opposite habitat. Since size distributions reduce each habitat into a common currency, we were able to examine the nature of the linkage. The relative response of each habitat to the nutrient treatment was different with regard to the shape of the distributions, while the relative response to the sediment treatment was different with regard to the biomass in the distributions.     

A principal-components approach based on heritability for combining phenotype information

For many traits, genetically relevant disease definition is unclear. For this reason, researchers applying linkage analysis often obtain information on a variety of items. With a large number of items, however, the test statistic from a multivariate analysis may require a prohibitively expensive correction for the multiple comparisons. The researcher is faced, therefore, with the issue of choosing which variables or combinations of variables to use in the linkage analysis. One approach to combining items is to first subject the data to a principal components analysis, and then perform the linkage analysis of the first few principal components. However, principal-components analyses do not take family structure into account. Here, an approach is developed in which family structure is taken into account when combining the data. The essence of the approach is to define principal components of heritability as the scores with maximum heritability in the data set, subject to being uncorrelated with each other. The principal components of heritability may be calculated as the solutions to a generalized eigensystem problem. Four simulation experiments are used to compare the power of linkage analyses based on the principal components of heritability and the usual principal components. The first of the experiments corresponds to the null hypothesis of no linkage. The second corresponds to a setting where the two kinds of principal components coincide. The third corresponds to a setting in which they are quite different and where the first of the usual principal components is not expected to have any power beyond the type I error rate. The fourth set of experiments corresponds to a setting where the usual principal components and the principal components of heritability differ, but where the first of the usual principal components is not without power. The results of the simulation experiments indicate that the principal components of heritability can be substantially different from the standard principal components and that when they are different, substantial gains in power can result by using the principal components of heritability in place of the standard principal components in linkage analyses.     

From components to regulatory motifs in signalling networks.

The developments in biochemistry and molecular biology over the past 30 years have produced an impressive parts list of cellular components. It has become increasingly clear that we need to understand how components come together to form systems. One area where this approach has been growing is cell signalling research. Here, instead of focusing on individual or small groups of signalling proteins, researchers are now using a more holistic perspective. This approach attempts to view how many components are working together in concert to process information and to orchestrate cellular phenotypic changes. Additionally, the advancements in experimental techniques to measure and visualize many cellular components at once gradually grow in diversity and accuracy. The multivariate data, produced by experiments, introduce new and exciting challenges for computational biologists, who develop models of cellular systems made up of interacting cellular components. The integration of high-throughput experimental results and information from legacy literature is expected to produce computational models that would rapidly enhance our understanding of the detail workings of mammalian cells.     

A high-throughput media design approach for high performance mammalian fed-batch cultures

An innovative high-throughput medium development method based on media blending was successfully used to improve the performance of a Chinese hamster ovary fed-batch medium in shaking 96-deepwell plates. Starting from a proprietary chemically-defined medium, 16 formulations testing 43 of 47 components at 3 different levels were designed. Media blending was performed following a custom-made mixture design of experiments considering binary blends, resulting in 376 different blends that were tested during both cell expansion and fed-batch production phases in one single experiment. Three approaches were chosen to provide the best output of the large amount of data obtained. A simple ranking of conditions was first used as a quick approach to select new formulations with promising features. Then, prediction of the best mixes was done to maximize both growth and titer using the Design Expert software. Finally, a multivariate analysis enabled identification of individual potential critical components for further optimization. Applying this high-throughput method on a fed-batch, rather than on a simple batch, process opens new perspectives for medium and feed development that enables identification of an optimized process in a short time frame.     

Comparing Trials with Multiple Outcomes: The Multivariate One-Sided Hypothesis with Unknown Covariances

The paper deals with a problem arising for tests in clinical trials. The outcomes of a standard and a new treatment to be compared are multivariate normally distributed with common but unknown covariance matrix. Under the null hypothesis the means of the outcomes are equal, under the alternative the new treatment is assumed to be superior, i.e. the means are larger without further quantification. For known covariance matrix there is a variety of tests for this problem. Some of these procedures can be extended to the case of unknown covariances if one is willing to accept a bias. There is, however, also an efficient unbiased test. The paper contains some numerical comparisons of these different procedures and takes a look on the minimax properties of the unbiased test.     

MorphoJ: an integrated software package for geometric morphometrics

Increasingly, data on shape are analysed in combination with molecular genetic or ecological information, so that tools for geometric morphometric analysis are required. Morphometric studies most often use the arrangements of morphological landmarks as the data source and extract shape information from them by Procrustes superimposition. The MorphoJ software combines this approach with a wide range of methods for shape analysis in different biological contexts. The program offers an integrated and user-friendly environment for standard multivariate analyses such as principal components, discriminant analysis and multivariate regression as well as specialized applications including phylogenetics, quantitative genetics and analyses of modularity in shape data. MorphoJ is written in Java and versions for the Windows, Macintosh and Unix/Linux platforms are freely available from http://www.flywings.org.uk/MorphoJ_page.htm.     

A mixed approach for proving non-inferiority in clinical trials with binary endpoints

When a new treatment is compared to an established one in a randomized clinical trial, it is standard practice to statistically test for non-inferiority rather than for superiority. When the endpoint is binary, one usually compares two treatments using either an odds-ratio or a difference of proportions. In this paper, we propose a mixed approach which uses both concepts. One first defines the non-inferiority margin using an odds-ratio and one ultimately proves non-inferiority statistically using a difference of proportions. The mixed approach is shown to be more powerful than the conventional odds-ratio approach when the efficacy of the established treatment is known (with good precision) and high (e.g. with more than 56% of success). The gain of power achieved may lead in turn to a substantial reduction in the sample size needed to prove non-inferiority. The mixed approach can be generalized to ordinal endpoints.     

A Hierarchical Bayesian Model for Spatial Prediction of Multivariate Non‐Gaussian Random Fields

Summary As most georeferenced data sets are multivariate and concern variables of different types, spatial mapping methods must be able to deal with such data. The main difficulties are the prediction of non‐Gaussian variables and the modeling of the dependence between processes. The aim of this article is to present a new hierarchical Bayesian approach that permits simultaneous modeling of dependent Gaussian, count, and ordinal spatial fields. This approach is based on spatial generalized linear mixed models. We use a moving average approach to model the spatial dependence between the processes. The method is first validated through a simulation study. We show that the multivariate model has better predictive abilities than the univariate one. Then the multivariate spatial hierarchical model is applied to a real data set collected in French Guiana to predict topsoil patterns.     

Accounting for probe-level noise in principal component analysis of microarray data

MOTIVATION: Principal Component Analysis (PCA) is one of the most popular dimensionality reduction techniques for the analysis of high-dimensional datasets. However, in its standard form, it does not take into account any error measures associated with the data points beyond a standard spherical noise. This indiscriminate nature provides one of its main weaknesses when applied to biological data with inherently large variability, such as expression levels measured with microarrays. Methods now exist for extracting credibility intervals from the probe-level analysis of cDNA and oligonucleotide microarray experiments. These credibility intervals are gene and experiment specific, and can be propagated through an appropriate probabilistic downstream analysis. RESULTS: We propose a new model-based approach to PCA that takes into account the variances associated with each gene in each experiment. We develop an efficient EM-algorithm to estimate the parameters of our new model. The model provides significantly better results than standard PCA, while remaining computationally reasonable. We show how the model can be used to 'denoise' a microarray dataset leading to improved expression profiles and tighter clustering across profiles. The probabilistic nature of the model means that the correct number of principal components is automatically obtained.     

Maximizing the Power of Principal-Component Analysis of Correlated Phenotypes in Genome-wide Association Studies

Many human traits are highly correlated. This correlation can be leveraged to improve the power of genetic association tests to identify markers associated with one or more of the traits. Principal component analysis (PCA) is a useful tool that has been widely used for the multivariate analysis of correlated variables. PCA is usually applied as a dimension reduction method: the few top principal components (PCs) explaining most of total trait variance are tested for association with a predictor of interest, and the remaining components are not analyzed. In this study we review the theoretical basis of PCA and describe the behavior of PCA when testing for association between a SNP and correlated traits. We then use simulation to compare the power of various PCA-based strategies when analyzing up to 100 correlated traits. We show that contrary to widespread practice, testing only the top PCs often has low power, whereas combining signal across all PCs can have greater power. This power gain is primarily due to increased power to detect genetic variants with opposite effects on positively correlated traits and variants that are exclusively associated with a single trait. Relative to other methods, the combined-PC approach has close to optimal power in all scenarios considered while offering more flexibility and more robustness to potential confounders. Finally, we apply the proposed PCA strategy to the genome-wide association study of five correlated coagulation traits where we identify two candidate SNPs that were not found by the standard approach.     

Estimating exposure effects by modelling the expectation of exposure conditional on confounders.

In order to estimate the causal effects of one or more exposures or treatments on an outcome of interest, one has to account for the effect of "confounding factors" which both covary with the exposures or treatments and are independent predictors of the outcome. In this paper we present regression methods which, in contrast to standard methods, adjust for the confounding effect of multiple continuous or discrete covariates by modelling the conditional expectation of the exposures or treatments given the confounders. In the special case of a univariate dichotomous exposure or treatment, this conditional expectation is identical to what Rosenbaum and Rubin have called the propensity score. They have also proposed methods to estimate causal effects by modelling the propensity score. Our methods generalize those of Rosenbaum and Rubin in several ways. First, our approach straightforwardly allows for multivariate exposures or treatments, each of which may be continuous, ordinal, or discrete. Second, even in the case of a single dichotomous exposure, our approach does not require subclassification or matching on the propensity score so that the potential for "residual confounding," i.e., bias, due to incomplete matching is avoided. Third, our approach allows a rather general formalization of the idea that it is better to use the "estimated propensity score" than the true propensity score even when the true score is known. The additional power of our approach derives from the fact that we assume the causal effects of the exposures or treatments can be described by the parametric component of a semiparametric regression model. To illustrate our methods, we reanalyze the effect of current cigarette smoking on the level of forced expiratory volume in one second in a cohort of 2,713 adult white males. We compare the results with those obtained using standard methods.     

A phenomics-based strategy identifies loci on APOC1, BRAP, and PLCG1 associated with metabolic syndrome phenotype domains

Despite evidence of the clustering of metabolic syndrome components, current approaches for identifying unifying genetic mechanisms typically evaluate clinical categories that do not provide adequate etiological information. Here, we used data from 19,486 European American and 6,287 African American Candidate Gene Association Resource Consortium participants to identify loci associated with the clustering of metabolic phenotypes. Six phenotype domains (atherogenic dyslipidemia, vascular dysfunction, vascular inflammation, pro-thrombotic state, central obesity, and elevated plasma glucose) encompassing 19 quantitative traits were examined. Principal components analysis was used to reduce the dimension of each domain such that >55% of the trait variance was represented within each domain. We then applied a statistically efficient and computational feasible multivariate approach that related eight principal components from the six domains to 250,000 imputed SNPs using an additive genetic model and including demographic covariates. In European Americans, we identified 606 genome-wide significant SNPs representing 19 loci. Many of these loci were associated with only one trait domain, were consistent with results in African Americans, and overlapped with published findings, for instance central obesity and FTO. However, our approach, which is applicable to any set of interval scale traits that is heritable and exhibits evidence of phenotypic clustering, identified three new loci in or near APOC1, BRAP, and PLCG1, which were associated with multiple phenotype domains. These pleiotropic loci may help characterize metabolic dysregulation and identify targets for intervention.     

A METHOD FOR ASSESSING PHYLOGENETIC LEAST SQUARES MODELS FOR SHAPE AND OTHER HIGH‐DIMENSIONAL MULTIVARIATE DATA

Studies of evolutionary correlations commonly use phylogenetic regression (i.e., independent contrasts and phylogenetic generalized least squares) to assess trait covariation in a phylogenetic context. However, while this approach is appropriate for evaluating trends in one or a few traits, it is incapable of assessing patterns in highly multivariate data, as the large number of variables relative to sample size prohibits parametric test statistics from being computed. This poses serious limitations for comparative biologists, who must either simplify how they quantify phenotypic traits, or alter the biological hypotheses they wish to examine. In this article, I propose a new statistical procedure for performing ANOVA and regression models in a phylogenetic context that can accommodate high‐dimensional datasets. The approach is derived from the statistical equivalency between parametric methods using covariance matrices and methods based on distance matrices. Using simulations under Brownian motion, I show that the method displays appropriate Type I error rates and statistical power, whereas standard parametric procedures have decreasing power as data dimensionality increases. As such, the new procedure provides a useful means of assessing trait covariation across a set of taxa related by a phylogeny, enabling macroevolutionary biologists to test hypotheses of adaptation, and phenotypic change in high‐dimensional datasets.