Search

This paper provides an update of the Learner Data Institute (LDI; www.learnerdatainstitute.org) which is now in its third year since conceptualization. Funded as a conceptualization project, the LDI's first two years had two major goals: (1) develop, implement, evaluate, and refine a framework for data-intensive science and engineering and (2) use the framework to start developing prototype solutions, based on data, data science, and science convergence, to a number of core challenges in learning science and engineering. One major focus in the third, current year is synthesizing efforts from the first two years to identify new opportunities for future research by various mutual interest groups within LDI, which have focused on developing a particular prototype solution to one or more related core challenges in learning science and engineering. In addition to highlighting emerging data-intensive solutions and innovations from the LDI's first two years, including places where LDI researchers have received additional funding for future research, we highlight here various core challenges our team has identified as being at a "tipping point." Tipping point challenges are those for which timely investment in data-intensive approaches has the maximum potential for a transformative effect.

Data reduction is crucial in order to turn large datasets into information, the major purpose of data science. The classic and richer area of dimensionality reduction (DR) has traditionally been based on feature extraction by combining primary features in a linear fashion, aiming to preserve or maintain covariance/correlations between the features. Nonlinear alternatives have been developed, including information-theoretic approaches using mutual information as well and conditional entropy based on target features. Here, we further this approach to feature selection or reduction strategy based on the concept of conditional Shannon entropy of two random variables. Novel results include (a) a dimensionality reduction method based on conditional entropy between predictors themselves along two variants, disregarding the influence of the target feature; (b) an error-prevention method inspired by error-detection and correction in information theory for DR with genomic data that can be used for abiotic data as well; and (c) a comparative assessment of the performance of several machine learning models on input features selected by these methods. We assess the quality of the techniques based on their performance in solving three application problems (Malware Classification, BioTaxonomy, and Noisy Classification) of various degrees of difficulty with competitive outcomes. Some useful heuristics arise from the analysis of the results and also suggest some problems of interest for further research.

Deep generative models with a specific variational autoencoding structure are capable of estimating parameters for the multidimensional logistic 2-parameter (ML2P) model in item response theory. In this work, we incorporated Q-matrix and variational autoencoder (VAE) to estimate item parameters with correlated and independent latent abilities, and we validate Q-matrix via the root mean square error (RMSE), bias, correlation, and AIC and BIC test score. The incorporation of a non-identity covariance matrix in a VAE requires a novel VAE architecture, which can be utilized in applications outside of education such as players performance evaluation, clinical trials assessment. Moreover, results show that the ML2P-VAE method is capable of estimating parameters and validating Q-matrix for models with a large number of latent variables with low computational cost, whereas traditional methods are infeasible for data with high-dimensional latent traits.

Earthquakes are one of the deadliest natural disasters. Our study focuses on detecting temporal patterns of earthquakes occurring along intraplate faults in the New Madrid seismic zone (NMSZ) within the middle of the United States from 1996–2016. Based on the magnitude and location of each earthquake, we developed a Bayesian clustering method to group hypocentres such that each group shared the same temporal pattern of occurrence. We constructed a matrix-variate Dirichlet process prior to describe temporal trends in the space and to detect regions showing similar temporal patterns. Simulations were conducted to assess accuracy and performance of the proposed method and to compare to other commonly used clustering methods such as Kmean, Kmedian and partition-around-medoids. We applied the method to NMSZ data to identify clusters of temporal patterns, which represent areas of stress that are potentially migrating over time. This information can then be used to assist in the prediction of future earthquakes.

We consider a general framework for constructing non-linear generators by adding a (32-bit or larger) pseudo-random number generator (PRNG) as a baseline generator to the basic RC4 design, in which an index-selection scheme similar to RC4 is used. We refer to the proposed design as the eRC (enhanced/extended RC4) design. We discuss several advantages of adding a good baseline generator to the RC4 design, including new updating schemes for the auxiliary table. We consider some popular PRNGs with the nice properties of high-dimensional equi-distribution, efficiency, long period, and portability as the baseline generator. We demonstrate that eRC generators are very efficient via extensive empirical testing on some eRC generators. We also show that eRC is flexible enough to choose minimal design parameters for eRC generators and yet the resulting eRC generators still pass stringent empirical tests, which makes them suitable for both software and hardware implementations.

Pseudo-random numbers (PRNs) are the basis for almost any statistical simulation and this depends largely on the quality of the pseudo-random number generator (PRNG) used. In this study, we used some results from number theory to propose an efficient method to accelerate the computer search of super-order maximum-period multiple recursive generators (MRGs). We conduct efficient computer searches and identify many efficient and portable MRGs of super-orders, 40751, 50551, and 50873; which respectively have equi-distribution property up to 40751, 50551, and 50873 dimensions, and period lengths of approximately 10380278.1, 10471730.6, and 10474729.3. Using the generalized Mersenne prime algorithm, we extend some known results of some efficient, portable and maximum-period MRGs. In particular, the DX/DL/DS/DT large-order generators are extended to super-order generators. An extensive empirical evaluation shows that these generators behave well when tested with the stringent Small Crush and Crush batteries of the TestU01 package.

Background Analyzing gene expression data rigorously requires taking assumptions into consideration but also relies on using information about network relations that exist among genes. Combining these different elements cannot only improve statistical power, but also provide a better framework through which gene expression can be properly analyzed. Material and methods We propose a novel statistical model that combines assumptions and gene network information into the analysis. Assumptions are important since every test statistic is valid only when required assumptions hold. So, we propose hybrid p-values and show that, under the null hypothesis of primary interest, these p-values are uniformly distributed. These proposed hybrid p-values take assumptions into consideration. We incorporate gene network information into the analysis because neighboring genes share biological functions. This correlation factor is taken into account via similar prior probabilities for neighboring genes. Results With a series of simulations our approach is compared with other approaches. Area Under the ROC Curves (AUCs) are constructed to compare the different methodologies; the AUC based on our methodology is larger than others. For regression analysis, AUC from our proposed method contains AUCs of Spearman test and of Pearson test. In addition, true negative rates (TNRs) also known as specificities are higher with our approach than with the other approaches. For two group comparison analysis, for instance, with a sample size of n=10, specificity corresponding to our proposed methodology is 0.716146 and specificities for t-test and rank sum are 0.689223 and 0.69797, respectively. Our method that combines assumptions and network information into the analysis is shown to be more powerful. Conclusions These proposed procedures are introduced as a general class of methods that can incorporate procedure-selection, account for multiple-testing, and incorporate graphical network information into the analysis. We obtain very good performance in simulations, and in real data analysis.

Epistemic network analysis (ENA) has been used in more than 300 published studies to date. However, there is no work in publication that describes the transformations that constitute ENA in formal mathematical terms. This paper provides such a description, focusing on the mathematical formulations that lead to two key affordances of ENA that are not present in other network analysis tools or multivariate analyses: (1) summary statistics that can be used to compare the differences in the content rather than the structure of networks and (2) network visualizations that provide information that is mathematically consistent with those statistics. Specifically, we describe the mathematical transformations by which ENA constructs matrix representations of discourse data, uses those representations to generate networks for units of analysis, places those networks into a metric space, identifies meaningful dimensions in the space, and positions the nodes of network graphs within that space so as to enable interpretation of those dimensions in terms of the content of the networks. We conclude with a discussion of how the mathematical formalisms of ENA can be used to model networks more generally.

Both vocalization and gesture are universal modes of communication and fundamental features of language development. Many believe that language evolved out of early gestural use; however, evidence reported here suggests vocalization precedes gesture in human communication and forms the predominant foundation for language. To our knowledge no prior research has investigated the rates of emergence of both gesture and vocalization in human infants to evaluate this question. We evaluated the rates of gesture and speech-like vocalizations (protophones) of 10 infants at 4, 7, and 11 months of age using parent-infant laboratory recordings. We found that infant protophones outnumbered gestures substantially at all three ages, ranging from >30 times more protophones than gestures at 3 months, to more than twice as many protophones as gestures at 11 months. The results suggest that vocalization is the predominant mode of communication in human infants from the beginning of life.

© This manuscript version is made available under the CC-BY-NC-ND 4.0 license https://creativecommons.org/licenses/by-nc-nd/4.0/. Deposited by shareyourpaper.org and openaccessbutton.org. We've taken reasonable steps to ensure this content doesn't violate copyright. However, if you think it does you can request a takedown by emailing help@openaccessbutton.org.

Pseudo-random number generators (PRNGs) play an important role in both areas of computer simulation and computer security. Currently, there appears to be a huge divide between the types of PRNGs used in these two areas. For PRNGs in computer security applications, the security concern is extremely important. For PRNGs in computer simulation applications, the properties of high-dimensional equi-distribution, efficiency, long period-length, and portability are important. In recent years, there have been many PRNGs proposed in the area of computer simulation satisfying these nice properties. However, most of them are linear generators, thus sharing the same weakness in predictability. The major aim of this article is to propose a general class of secure generators, called SAFE (secure and fast encryption) generators, by properly “mixing” two baseline generators with the aforementioned properties to obtain a secure generator that would inherit these nice properties. Specifically, we propose applying a general mutual-shuffling method to certain linear generators, such as the currently most popular MT19937 generator and large-order multiple recursive generators, as well as outputting certain nonlinear transformations of the generated variates to construct secure PRNGS.

Both vocalization and gesture are universal modes of communication and fundamental features of language development. The gestural origins theory proposes that language evolved out of early gestural use. However, evidence reported here suggests vocalization is much more prominent in early human communication than gesture is. To our knowledge no prior research has investigated the rates of emergence of both gesture and vocalization across the first year in human infants. We evaluated the rates of gestures and speech-like vocalizations (protophones) in 10 infants at 4, 7, and 11 months of age using parent-infant laboratory recordings. We found that infant protophones outnumbered gestures substantially at all three ages, ranging from >35 times more protophones than gestures at 3 months, to >2.5 times more protophones than gestures at 11 months. The results suggest vocalization, not gesture, is the predominant mode of communication in human infants in the first year.

Urban forests and parks are important for recreation and may serve as a natural corridor for commuters. The consequences of human-mediated disturbance in natural areas are documented for avian and mammalian species. Less is known about the consequences of human disturbance on reptile species, specifically snakes, residing in natural refuges within the urban matrix. Thus, we examined the spatial activity of copperheads (Agkistrodon contortrix) in regard to pedestrian trails within an urban forest. We used radio telemetry to track snakes during the active season and estimated distances moved in between relocations, distances to the nearest trail and home range size for individuals. We found sex and season, but not distance to the nearest trail, affected the distance snakes moved. In addition, we observed a weak, positive relationship between home range size and average distance to the trail. Sex, season and body condition did not explain snake distance to the trail, but individual patterns were variable for snakes compared to random locations generated from snake relocations. Our study indicates copperheads may be tolerant of low-level human disturbances found in an urban forest. Further work should be done to quantify levels of disturbance, such as trail use, and compare the behavior of reptiles across urban park types and locations.

Research on infant vocal development has provided notable insights into vocal interaction with caregivers, elucidating growth in foundations for language through parental elicitation and reaction to vocalizations. A role for infant vocalizations produced endogenously, potentially providing raw material for interaction and a basis for growth in the vocal capacity itself, has received less attention. We report that in laboratory recordings of infants and their parents, the bulk of infant speech-like vocalizations, or “protophones”, were directed toward no one and instead appeared to be generated endogenously, mostly in exploration of vocal abilities. The tendency to predominantly produce protophones without directing them to others occurred both during periods when parents were instructed to interact with their infantsandduring periods when parents were occupied with an interviewer, with the infants in the room. The results emphasize the infant as an agent in vocal learning, even when not interacting socially and suggest an enhanced perspective on foundations for vocal language.

Temporal coordination of vocal exchanges between mothers and their infants emerges from a developmental process that relies on the ability of communication partners to co-coordinate and predict each other's turns. Consequently, the partners engage in communicative niche construction that forms a foundation for language in human infancy. While robust universals in vocal turn-taking have been found, differences in the timing of maternal and infant vocalizations have also been reported across cultures. In this study, we examine the temporal structure of vocal interactions in 38 mother-infant dyads in the first two years across two cultures-American and Lebanese-by studying observed and randomized distributions of vocalizations, focusing on both gaps and overlaps in naturalistic 10-min vocal interactions. We conducted a series of simulations using Kolmogorov-Smirnov (K-S) tests to examine whether the observed responsivity patterns differed from randomly generated simulations of responsivity patterns in both Arabic and English for mothers responding to infants and for infants responding to mothers. Results revealed that both mothers and infants engaged in conversational alternation, with mothers acting similarly across cultures. By contrast, significant differences were observed in the timing of infant responses to maternal utterances, with the Lebanese infants' tendency to cluster their responses in the first half-second after the offset of the Lebanese mothers' utterances to a greater extent than their American counterparts. We speculate that the results may be due to potential phonotactic differences between Arabic and English and/or to differing child-rearing practices across Lebanese and American cultures. The findings may have implications for early identification of developmental disorders such as autism spectrum disorders within and across cultures.

How did vocal language originate? Before trying to determine how referential vocabulary or syntax may have arisen, it is critical to explain how ancient hominins began to produce vocalization flexibly, without binding to emotions or functions. A crucial factor in the vocal communicative split of hominins from the ape background may thus have been copious, functionally flexible vocalization, starting in infancy and continuing throughout life, long before there were more advanced linguistic features such as referential vocabulary. 2–3 month-old modern human infants produce “protophones”, including at least three types of functionally flexible non-cry precursors to speech rarely reported in other ape infants. But how early in life do protophones actually appear? We report that the most common protophone types emerge abundantly as early as vocalization can be observed in infancy, in preterm infants still in neonatal intensive care. Contrary to the expectation that cries are the predominant vocalizations of infancy, our all-day recordings showed that protophones occurred far more frequently than cries in both preterm and full-term infants. Protophones were not limited to interactive circumstances, but also occurred at high rates when infants were alone, indicating an endogenous inclination to vocalize exploratorily, perhaps the most fundamental capacity underlying vocal language.

Aim: Phytochemicals from fruits and vegetables are known to reduce inflammation and improve overall health. The objective of this study was to determine the effect of a fruit and vegetable concentrate (FVC) and high fiber component on the gut microbiome in an overweight/obese, female population. Methods: The study was a randomized, double blind, placebo-controlled trial with 57 asymptomatic, pre-menopausal, overweight/obese females between 25–50 years of age working in healthcare. Blood and fecal samples were collected before and after two, four and five months of daily supplementation. Metabolic parameters were measured, and the gut microbiome analyzed. Results: No effect was observed with FVC supplementation for blood lipids, glucose and immune parameters. There was an improvement in glucose clearance. The FVC supplement did not result in taxonomic alterations at phyla level, or changes in α or β diversity, but reduced Bacteroides abundance and increased fecal butyrate. An additional high fiber component improved levels of health associated bacteria. Conclusion: The results suggest that a dried fruit and vegetable supplement, with a high fiber meal replacement can alter the intestinal microbiota and improve glucose clearance, suggesting that this combination of supplements can improve glucose metabolism and possibly reduce the risk of insulin resistance.

Many study designs yield a variety of outcomes from each subject clustered within an experimental unit. When these outcomes are of mixed data types, it is challenging to jointly model the e...

Binary data often occur in epidemiology studies clustered by families. When an assumption of exchangeability of response is reasonable, the exchangeable multinomial distribution may be used to model binary and multinomial responses. Conditional probabilities of a new family member being diagnosed given family members previously diagnosed are derived. Applications are made to real familial disease data on IPF and ALS.

Parametric models of discrete data with exchangeable dependence structure present substantial computational challenges for maximum likelihood estimation. Coordinate descent algorithms such as the Newton’s method are usually unstable, becoming a hit or miss adventure on initialization with a good starting value. We propose a method for computing maximum likelihood estimates of parametric models for finitely exchangeable binary data, formalized as an iterative weighted least squares algorithm.

In this article, maximum likelihood estimates of an exchangeable multinomial distribution using a parametric form to model the parameters as functions of covariates are derived. The non linearity o...

OBJECTIVES: Phytochemicals from fruits and vegetables have the potential to reduce inflammation and improve overall health. This study evaluated the effect of an encapsulated fruit and vegetable juice concentrate (FVC) on the composition and abundances of intestinal microbiota in an overweight/obese, female population and the association with indicators of metabolic health and inflammatory status. METHODS: The study was performed as a randomized, double blind, placebo-controlled trial with 57 asymptomatic, pre-menopausal, overweight/obese (BMI between 25 and 40 kg/m(2)) females between 25–50 years of age and working in the healthcare field. Blood and fecal samples were collected before and after eight and sixteen weeks of daily supplementation with FVC. Metabolic and immune parameters were measured, and the gut microbiome was analyzed by sequencing of the V4 region of the 16SrRNA gene. RESULTS: The FVC supplementation did not change serum cholesterol, HDL, LDL, triglycerides and fasting glucose, or fasting cytokine levels for MCP1, MIP-1β, TNF-α, IL-10, IL-1β, IL-6, and Leptin. Sequencing analysis revealed the most abundant phyla detected at baseline in this population were Firmicutes (75.75 ± 10.59%), Bacteroidetes (15.23 ± 12.5%), with Actinobacteria, Proteobacteria and Verrucomicrobia also present. The following genera was most abundant in the study population: Blautia (22 ± 12.86%), Roseburia (8.03 ± 7.34%), Coprococus (5.44 ± 4.0%), Faecalibatcerium (11.12 ± 8.12%) and Bacteroides (10.89 ± 9.56%). FCV supplementation did not result in significant taxonomic changes at the level of phyla, or alter alpha (Shannon Index) or beta diversity (Bray Curtis analysis). However, there was a consistent reduction in overall bacterial abundance in the FVC supplemented group. This reduction was primarily attributed to the decrease in the gram-negative, anaerobic genus, Bacteroides (P = 0.08). CONCLUSIONS: An encapsulated fruit and vegetable juice concentrate did not alter metabolic or inflammatory indicators in a healthy, overweight/obese female population, but resulted in a reduced abundance of the intestinal bacterial population, Bacteroides. FUNDING SOURCES: NSA LLC/The Juice Plus(+®) Company.

Background It has been a challenging task to build a genome-wide phylogenetic tree for a large group of species containing a large number of genes with long nucleotides sequences. The most popular method, called feature frequency profile (FFP-k), finds the frequency distribution for all words of certain length k over the whole genome sequence using (overlapping) windows of the same length. For a satisfactory result, the recommended word length (k) ranges from 6 to 15 and it may not be a multiple of 3 (codon length). The total number of possible words needed for FFP-k can range from 46=4096 to 415. Results We propose a simple improvement over the popular FFP method using only a typical word length of 3. A new method, called Trinucleotide Usage Profile (TUP), is proposed based only on the (relative) frequency distribution using non-overlapping windows of length 3. The total number of possible words needed for TUP is 43=64, which is much less than the total count for the recommended optimal “resolution” for FFP. To build a phylogenetic tree, we propose first representing each of the species by a TUP vector and then using an appropriate distance measure between pairs of the TUP vectors for the tree construction. In particular, we propose summarizing a DNA sequence by a matrix of three rows corresponding to three reading frames, recording the frequency distribution of the non-overlapping words of length 3 in each of the reading frame. We also provide a numerical measure for comparing trees constructed with various methods. Conclusions Compared to the FFP method, our empirical study showed that the proposed TUP method is more capable of building phylogenetic trees with a stronger biological support. We further provide some justifications on this from the information theory viewpoint. Unlike the FFP method, the TUP method takes the advantage that the starting of the first reading frame is (usually) known. Without this information, the FFP method could only rely on the frequency distribution of overlapping words, which is the average (or mixture) of the frequency distributions of three possible reading frames. Consequently, we show (from the entropy viewpoint) that the FFP procedure could dilute important gene information and therefore provides less accurate classification.

This paper develops threshold models for developmental toxicity data. The distinguishing feature of these threshold models is their flexibility in modeling data below threshold with a U-shaped function if the data warrants. The method is applied to actual data from a developmental study which exhibits U-shaped behavior in early dose groups. Results from a simulation study demonstrate the flexibility of the threshold model to pick up on U-shaped trends in the data. In addition, the simulation study reveals important considerations in design of developmental studies.

Estimates of mean response for a developmental toxicity study are developed using the techniques of Bayesian bootstrap. Using this method, a joint posterior distribution of mean response is simulated, providing a means for determining estimated variance and confidence statements. The approach allows for effects on litter size to be taken into consideration in the estimation of mean response. In addition a method is given for the incorporation of prior information into the analysis. The prior information may be information about mean response and about the litter size distribution as well. Results are compared with likelihood based estimates.

Monte Carlo simulations have become a common practice to evaluate a proposed statistical procedure, particularly when it is analytically intractable. Validity of any simulation study relies heavily on the goodness of random variate generators for some specified distributions, which in turn is based on the successful generation of independent variates from the uniform distribution. However, a typical computer-generated pseudo-random number generator (PRNG) is a deterministic algorithm and we know that no PRNG is capable of generating a truly random uniform sequence. Since the foundation of a simulation study is built on the PRNG used, it is extremely important to design a good PRNG. We review some recent developments on PRNGs with nice properties such as high-dimensional equi-distribution, efficiency, long period length, portability, and efficient parallel implementations. For further resources related to this article, please visit the WIREs website.

A simulation study is conducted to determine the effects of varying correlation structures on two estimation procedures used to model clustered binary data; a parametric model, the beta-binomial, and a non-parametric model, the exchangeable binary. The simulations detected bias in estimation of the mean response parameter and the correlation parameter when assuming a parametric model. In addition it was found that variance parameters can be severely underestimated if the correlation structure is considered strictly a nuisance parameter.

This paper proposes an independence test for a set of clustered binary observations, such as might be encountered in developmental toxicity studies. An exchangeable binary model is employed, under an assumption of exchangeability among cluster elements, to model probability of positive response. With a Weibull form assumed for response, the independence test is equivalent to testing whether a parameter value is unity. This Weibull form also allows for parametric tests for a covariate-response relationship and for covariate effects on correlation. The procedure is illustrated using data obtained from a developmental toxicity study.

Correlated binary data occur very frequently in statistical practice. In many applications, it is reasonable to assume that data from the same cluster are exchangeable. Such data are commonly encountered in cluster sample surveys, teratological experiments, ophthalmologic and otolaryngologic studies, and other clinical trials. The standard methods of analyzing these data include the use of beta-binomial models and generalized estimating equations with third and fourth moments specified by “working matrices.” The focus of these procedures is an estimation of the mean and variance parameters. More information can be obtained when data are exchangeable. By expressing the joint distribution of a set of exchangeable binary random variables in terms of the probability of similar response within cluster, this article introduces a procedure for obtaining maximum likelihood estimates of population parameters such as the marginal means, moments, and correlations of orders two and higher. Applications are m...

Hormetic effects have been observed at low exposure levels based on the dose-response pattern of data from developmental toxicity studies. This indicates that there might actually be a reduced risk of exhibiting toxic effects at low exposure levels. Hormesis implies the existence of a threshold dose level and there are dose-response models that include parameters that account for the threshold. We propose a function that introduces a parameter to account for hormesis. This function is a subset of the set of all functions that could represent a hormetic dose-response relationship at low exposure levels to toxic agents. We characterize the overall dose-response relationship with a piecewise function that consists of a hormetic u-shape curve at low dose levels and a logistic curve at high dose levels. We apply our model to a data set from an experiment conducted at the National Toxicology Program (NTP). We also use the beta-binomial distribution to model the litter response data. It can be seen by observing the structure of these data that current experimental designs for developmental studies employ a limited number of dose groups. These designs may not be satisfactory when the goal is to illustrate the existence of hormesis. In particular, increasing the number of low-level doses improves the power for detecting hormetic effects. Therefore, we also provide the results of simulations that were done to characterize the power of current designs in detecting hormesis and to demonstrate how this power can be improved upon by altering these designs with the addition of only a few low exposure levels.

A tiered testing strategy has been developed to evaluate the potential of tobacco processes, ingredients, or technological developments to change the biological activity resulting from burning tobacco. The strategy is based on comparative chemical and biological testing. Dry ice expanded tobacco (DIET) is an example of a common tobacco expansion process currently used in the manufacture of cigarettes to increase tobacco filling capacity. As part of the toxicological evaluation of DIET, test cigarettes containing DIET were compared with control cigarettes containing tobacco expanded with a traditional expansion agent (Freon-11, also known as trichlorofluoromethane). Testing included mainstream cigarette smoke chemistry studies, genotoxicity studies (Ames and sister chromatid exchange, SCE), a 13-week inhalation study in Sprague-Dawley rats, and a 30-week dermal tumor promotion study in SENCAR mice. Cigarettes containing DIET or Freon-11 expanded tobacco were similar in biological activity.

Agency history, basic principles, and attitudes influence a comprehensive version of psychiatric rehabilitation.

The presence of intralitter correlation is a well known issue for analysis of the developmental toxicology data. The intralitter correlation coefficients observed in developmental toxicology data are generally different across dose groups. In this paper we use a generalized estimating equation procedure to model jointly the mean parameters and the intralitter correlation coefficients as functions of dose levels. Our procedure is similar to that used by Prentice and Zhao (1991, Biometrics 47, 825-839) for estimating the mean and variance parameters.

Background The dimension and complexity of high-throughput gene expression data create many challenges for downstream analysis. Several approaches exist to reduce the number of variables with respect to small sample sizes. In this study, we utilized the Generalized Double Pareto (GDP) prior to induce sparsity in a Bayesian Generalized Linear Model (GLM) setting. The approach was evaluated using a publicly available microarray dataset containing 99 samples corresponding to four different prostate cancer subtypes. Results A hierarchical Sparse Bayesian GLM using GDP prior (SBGG) was developed to take into account the progressive nature of the response variable. We obtained an average overall classification accuracy between 82.5% and 94%, which was higher than Support Vector Machine, Random Forest or a Sparse Bayesian GLM using double exponential priors. Additionally, SBGG outperforms the other 3 methods in correctly identifying pre-metastatic stages of cancer progression, which can prove extremely valuable for therapeutic and diagnostic purposes. Importantly, using Geneset Cohesion Analysis Tool, we found that the top 100 genes produced by SBGG had an average functional cohesion p-value of 2.0E-4 compared to 0.007 to 0.131 produced by the other methods. Conclusions Using GDP in a Bayesian GLM model applied to cancer progression data results in better subclass prediction. In particular, the method identifies pre-metastatic stages of prostate cancer with substantially better accuracy and produces more functionally relevant gene sets.