Rat toxicogenomic study reveals analytical consistency across microarray platforms

Discovery of molecular mechanisms of traditional Chinese medicinal formula Si-Wu-Tang using gene expression microarray and connectivity map

To pursue a systematic approach to discovery of mechanisms of action of traditional Chinese medicine (TCM), we used microarrays, bioinformatics and the “Connectivity Map” (CMAP) to examine TCM-induced changes in gene expression. We demonstrated that this approach can be used to elucidate new molecular targets using a model TCM herbal formula Si-Wu-Tang (SWT) which is widely used for women's health. The human breast cancer MCF-7 cells treated with 0.1 µM estradiol or 2.56 mg/ml of SWT showed dramatic gene expression changes, while no significant change was detected for ferulic acid, a known bioactive compound of SWT. Pathway analysis using differentially expressed genes related to the treatment effect identified that expression of genes in the nuclear factor erythroid 2-related factor 2 (Nrf2) cytoprotective pathway was most significantly affected by SWT, but not by estradiol or ferulic acid. The Nrf2-regulated genes HMOX1, GCLC, GCLM, SLC7A11 and NQO1 were upreguated by SWT in a dose-dependent manner, which was validated by real-time RT-PCR. Consistently, treatment with SWT and its four herbal ingredients resulted in an increased antioxidant response element (ARE)-luciferase reporter activity in MCF-7 and HEK293 cells. Furthermore, the gene expression profile of differentially expressed genes related to SWT treatment was used to compare with those of 1,309 compounds in the CMAP database. The CMAP profiles of estradiol-treated MCF-7 cells showed an excellent match with SWT treatment, consistent with SWT's widely claimed use for women's diseases and indicating a phytoestrogenic effect. The CMAP profiles of chemopreventive agents withaferin A and resveratrol also showed high similarity to the profiles of SWT. This study identified SWT as an Nrf2 activator and phytoestrogen, suggesting its use as a nontoxic chemopreventive agent, and demonstrated the feasibility of combining microarray gene expression profiling with CMAP mining to discover mechanisms of actions and to identify new health benefits of TCMs.

The vitamin D receptor and inducible nitric oxide synthase associated pathways in acquired resistance to Cooperia oncophora infection in cattle

Cooperia oncophora is an economically important gastrointestinal nematode in ruminants. Acquired resistance to Cooperia oncophora infection in cattle develops rapidly as a result of prior infections. Naïve cattle, when given a primary infection of high-dose infective L3 larvae, develop a strong immunity to subsequent reinfection. Compared to primary infection, reinfection resulted in a marked reduction in worm establishment. In order to understand molecular mechanisms underlying the development of acquired resistance, we characterized the transcriptomic responses of the bovine small intestine to a primary infection and reinfection. A total of 23 pathways were significantly impacted during infection. The vitamin D receptor activation was strongly induced only during reinfection, suggesting that this pathway may play an important role in the development of acquired resistance via its potential roles in immune regulation and intestinal mucosal integrity maintenance. The expression of inducible nitric oxide synthase (NOS2) was strongly induced during reinfection but not during primary infection. As a result, several canonical pathways associated with NOS2 were impacted. The genes involved in eicosanoid synthesis, including prostaglandin synthase 2 (PTGS2 or COX2), remained largely unchanged during infection. The rapid development of acquired resistance may help explain the lack of relative pathogenicity by Cooperia oncophora infection in cattle. Our findings facilitate the understanding of molecular mechanisms underlying the development of acquired resistance, which could have an important implication in vaccine design.

Serious complications in gene-expression studies with stress perturbation: An example of UV-exposed p53-mutant mouse embryonic fibroblasts

Reanalysis of our UV study of p53-mutant mouse embryonic fibroblasts revealed an intriguing orchestration of massive transcriptome responses. However, close scrutiny of the data uncovered an affected mRNA/rRNA ratio, effectively inhibiting valid data analysis. UV-dose range-finding showed low-dose UV specific- and high-dose stress-related responses, which represent a plea for UV dose range-finding in experimental design.

A possible mechanism for hepatotoxicity induced by BIRB-796, an orally active p38 mitogen-activated protein kinase inhibitor

BIRB-796, a selective inhibitor of p38 mitogen-activated protein kinase, has entered clinical trials for the treatment of autoimmune diseases. Levels of alanine transaminase, a biomarker of hepatic toxicity in clinical pathology, were found to be increased in Crohn's disease patients treated with BIRB-796. The purpose of the present study was to clarify the molecular mechanism(s) of this hepatotoxicity. A toxicogenomic analysis using a highly sensitive DNA chip, 3D-Gene™ Mouse Oligo chip 24k, indicated that BIRB-796 treatment activated the nuclear factor (erythroid-derived 2)-like 2 signaling pathway, which plays a key role in the response to oxidative stress. A reactive intermediate of BIRB-796 was detected by the glutathione-trapping method using mouse and human liver microsomes. The production of this reactive metabolite in the liver may be one of the causes of BIRB-796's hepatotoxicity.

Functional modules with disease discrimination abilities for various cancers

Selecting differentially expressed genes (DEGs) is one of the most important tasks in microarray applications for studying multi-factor diseases including cancers. However, the small samples typically used in current microarray studies may only partially reflect the widely altered gene expressions in complex diseases, which would introduce low reproducibility of gene lists selected by statistical methods. Here, by analyzing seven cancer datasets, we showed that, in each cancer, a wide range of functional modules have altered gene expressions and thus have high disease classification abilities. The results also showed that seven modules are shared across diverse cancers, suggesting hints about the common mechanisms of cancers. Therefore, instead of relying on a few individual genes whose selection is hardly reproducible in current microarray experiments, we may use functional modules as functional signatures to study core mechanisms of cancers and build robust diagnostic classifiers.

Cross-platform comparison of microarray data using order restricted inference

MOTIVATION

Titration experiments measuring the gene expression from two different tissues, along with total RNA mixtures of the pure samples, are frequently used for quality evaluation of microarray technologies. Such a design implies that the true mRNA expression of each gene, is either constant or follows a monotonic trend between the mixtures, applying itself to the use of order restricted inference procedures. Exploiting only the postulated monotonicity of titration designs, we propose three statistical analysis methods for the validation of high-throughput genetic data and corresponding preprocessing techniques.

RESULTS

Our methods allow for inference of accuracy, repeatability and cross-platform agreement, with minimal required assumptions regarding the underlying data generating process. Therefore, they are readily applicable to all sorts of genetic high-throughput data independent of the degree of preprocessing. An application to the EMERALD dataset was used to demonstrate how our methods provide a rich spectrum of easily interpretable quality metrics and allow the comparison of different microarray technologies and normalization methods. The results are on par with previous work, but provide additional new insights that cast doubt on the utility of popular preprocessing techniques, specifically concerning the EMERALD projects dataset.

AVAILABILITY

All datasets are available on EBI's ArrayExpress web site http://www.ebi.ac.uk/microarray-as/ae/) under accession numbers E-TABM-536, E-TABM-554 and E-TABM-555. Source code implemented in C and R is available at: http://statistics.msi.meduniwien.ac.at/float/cross_platform/. Methods for testing and variance decomposition have been made available in the R-package orQA, which can be downloaded and installed from CRAN http://cran.r-project.org.

Utilizing toxicogenomic data to understand chemical mechanism of action in risk assessment

The predominant role of toxicogenomic data in risk assessment, thus far, has been one of augmentation of more traditional in vitro and in vivo toxicology data. This article focuses on the current available examples of instances where toxicogenomic data has been evaluated in human health risk assessment (e.g., acetochlor and arsenicals) which have been limited to the application of toxicogenomic data to inform mechanism of action. This article reviews the regulatory policy backdrop and highlights important efforts to ultimately achieve regulatory acceptance. A number of research efforts on specific chemicals that were designed for risk assessment purposes have employed mechanism or mode of action hypothesis testing and generating strategies. The strides made by large scale efforts to utilize toxicogenomic data in screening, testing, and risk assessment are also discussed. These efforts include both the refinement of methodologies for performing toxicogenomics studies and analysis of the resultant data sets. The current issues limiting the application of toxicogenomics to define mode or mechanism of action in risk assessment are discussed together with interrelated research needs. In summary, as chemical risk assessment moves away from a single mechanism of action approach toward a toxicity pathway-based paradigm, we envision that toxicogenomic data from multiple technologies (e.g., proteomics, metabolomics, transcriptomics, supportive RT-PCR studies) can be used in conjunction with one another to understand the complexities of multiple, and possibly interacting, pathways affected by chemicals which will impact human health risk assessment.

Cross-platform comparison of microarray-based multiple-class prediction

High-throughput microarray technology has been widely applied in biological and medical decision-making research during the past decade. However, the diversity of platforms has made it a challenge to re-use and/or integrate datasets generated in different experiments or labs for constructing array-based diagnostic models. Using large toxicogenomics datasets generated using both Affymetrix and Agilent microarray platforms, we carried out a benchmark evaluation of cross-platform consistency in multiple-class prediction using three widely-used machine learning algorithms. After an initial assessment of model performance on different platforms, we evaluated whether predictive signature features selected in one platform could be directly used to train a model in the other platform and whether predictive models trained using data from one platform could predict datasets profiled using the other platform with comparable performance. Our results established that it is possible to successfully apply multiple-class prediction models across different commercial microarray platforms, offering a number of important benefits such as accelerating the possible translation of biomarkers identified with microarrays to clinically-validated assays. However, this investigation focuses on a technical platform comparison and is actually only the beginning of exploring cross-platform consistency. Further studies are needed to confirm the feasibility of microarray-based cross-platform prediction, especially using independent datasets.

Transcriptome analysis of common gene expression in human mesenchymal stem cells derived from four different origins

We have used Affymetrix oligonucleotide microarrays to analyze common transcriptomes and thereby learn about the core gene expression profile in human mesenchymal stem cells (MSC) from different tissues, including fetal amniotic fluid-derived MSC, term pregnancy amniotic membrane-derived MSC, term pregnancy umbilical cord blood-derived MSC, and adult bone marrow-derived MSC. The beauty of microarray analysis of gene expression (MAGE) is that it can be used to discover associating genes that were previously thought to be unrelated to a physiological or pathological event. However, interpreting complex biological processes from gene expression profiles often requires extensive knowledge mining in biomedical literature. In this chapter, we describe, step-by-step, how to use a commercially available biological database and software program, MetaCore (GeneGo Inc.), for functional network analysis.

Comparison of Beta-value and M-value methods for quantifying methylation levels by microarray analysis

Background

High-throughput profiling of DNA methylation status of CpG islands is crucial to understand the epigenetic regulation of genes. The microarray-based Infinium methylation assay by Illumina is one platform for low-cost high-throughput methylation profiling. Both Beta-value and M-value statistics have been used as metrics to measure methylation levels. However, there are no detailed studies of their relations and their strengths and limitations.

Results

We demonstrate that the relationship between the Beta-value and M-value methods is a Logit transformation, and show that the Beta-value method has severe heteroscedasticity for highly methylated or unmethylated CpG sites. In order to evaluate the performance of the Beta-value and M-value methods for identifying differentially methylated CpG sites, we designed a methylation titration experiment. The evaluation results show that the M-value method provides much better performance in terms of Detection Rate (DR) and True Positive Rate (TPR) for both highly methylated and unmethylated CpG sites. Imposing a minimum threshold of difference can improve the performance of the M-value method but not the Beta-value method. We also provide guidance for how to select the threshold of methylation differences.

Conclusions

The Beta-value has a more intuitive biological interpretation, but the M-value is more statistically valid for the differential analysis of methylation levels. Therefore, we recommend using the M-value method for conducting differential methylation analysis and including the Beta-value statistics when reporting the results to investigators.

Gene expression profiling for discovery of novel targets in human traumatic brain injury

Several clinical trials have failed to demonstrate a significant effect on outcome following human traumatic brain injury (TBI) despite promising results obtained in preclinical animal studies. These failures may be due in part to a misinterpretation of the findings obtained in preclinical animal models of TBI, a misunderstanding of the complexity of the human response to TBI, limited knowledge about the biological pathways that interact to contribute to good and bad outcomes after brain injury, and the effects of genomic variability and environment on individual recovery. Recent publications suggest that data obtained from gene expression profiling studies of complex neurological diseases such as stroke, multiple sclerosis (MS), Alzheimer's and Parkinson's may contribute to a more informed understanding of what affects outcome following TBI. These data may help to bridge the gap between successful preclinical studies and negative clinical trials in humans to reveal novel targets for therapy. Gene expression profiling has the capability to identify biomarkers associated with response to TBI, elucidate complex genetic interactions that may play a role in outcome following TBI, and reveal biological pathways related to brain health. This review highlights the current state of the literature on gene expression profiling for neurological disease and discusses its ability to aid in unraveling the variable human response to TBI and the potential for it to offer treatment strategies in an area where we currently have limited therapeutic options primarily based on supportive care.

Evaluation of external RNA controls for the standardisation of gene expression biomarker measurements

Background

Gene expression profiling is an important approach for detecting diagnostic and prognostic biomarkers, and predicting drug safety. The development of a wide range of technologies and platforms for measuring mRNA expression makes the evaluation and standardization of transcriptomic data problematic due to differences in protocols, data processing and analysis methods. Thus, universal RNA standards, such as those developed by the External RNA Controls Consortium (ERCC), are proposed to aid validation of research findings from diverse platforms such as microarrays and RT-qPCR, and play a role in quality control (QC) processes as transcriptomic profiling becomes more commonplace in the clinical setting.

Results

Panels of ERCC RNA standards were constructed in order to test the utility of these reference materials (RMs) for performance characterization of two selected gene expression platforms, and for discrimination of biomarker profiles between groups. The linear range, limits of detection and reproducibility of microarray and RT-qPCR measurements were evaluated using panels of RNA standards. Transcripts of low abundance (≤ 10 copies/ng total RNA) showed more than double the technical variability compared to higher copy number transcripts on both platforms. Microarray profiling of two simulated 'normal' and 'disease' panels, each consisting of eight different RNA standards, yielded robust discrimination between the panels and between standards with varying fold change ratios, showing no systematic effects due to different labelling and hybridization runs. Also, comparison of microarray and RT-qPCR data for fold changes showed agreement for the two platforms.

Conclusions

ERCC RNA standards provide a generic means of evaluating different aspects of platform performance, and can provide information on the technical variation associated with quantification of biomarkers expressed at different levels of physiological abundance. Distinct panels of standards serve as an ideal quality control tool kit for determining the accuracy of fold change cut-off threshold and the impact of experimentally-derived noise on the discrimination of normal and disease profiles.

Consistency of predictive signature genes and classifiers generated using different microarray platforms

Microarray-based classifiers and associated signature genes generated from various platforms are abundantly reported in the literature; however, the utility of the classifiers and signature genes in cross-platform prediction applications remains largely uncertain. As part of the MicroArray Quality Control Phase II (MAQC-II) project, we show in this study 80-90% cross-platform prediction consistency using a large toxicogenomics data set by illustrating that: (1) the signature genes of a classifier generated from one platform can be directly applied to another platform to develop a predictive classifier; (2) a classifier developed using data generated from one platform can accurately predict samples that were profiled using a different platform. The results suggest the potential utility of using published signature genes in cross-platform applications and the possible adoption of the published classifiers for a variety of applications. The study reveals an opportunity for possible translation of biomarkers identified using microarrays to clinically validated non-array gene expression assays.

Clinical application of gene expression profiling in breast cancer

Breast cancer is a heterogeneous disease associated with variable clinical outcomes and response to therapy. Classic clinicopathologic factors associated with outcome include anatomic features associated with prognosis (eg, tumor size, number of positive regional lymph nodes) and biologic features associated with prognosis and/or predictive of response to specific therapies, usually by evaluating protein expression by immunohistochemistry (eg, estrogen and/or progesterone receptors) or amplification of a single gene (eg, HER2/neu). Gene expression profiling evaluating thousands of genes is now feasible, and has facilitated the development of multiparameter assays that may identify breast cancer subtypes associated with distinct clinical outcomes that were not previously recognized, or provide more accurate information about prognosis or response to specific therapies than may be provided by classic clinicopathologic features alone. Several multiparameter gene expression assays are commercially available, and additional assays are being developed that will facilitate more accurate therapeutic individualization.

Evaluation of gene expression data generated from expired Affymetrix GeneChip® microarrays using MAQC reference RNA samples

BACKGROUND

The Affymetrix GeneChip® system is a commonly used platform for microarray analysis but the technology is inherently expensive. Unfortunately, changes in experimental planning and execution, such as the unavailability of previously anticipated samples or a shift in research focus, may render significant numbers of pre-purchased GeneChip® microarrays unprocessed before their manufacturer's expiration dates. Researchers and microarray core facilities wonder whether expired microarrays are still useful for gene expression analysis. In addition, it was not clear whether the two human reference RNA samples established by the MAQC project in 2005 still maintained their transcriptome integrity over a period of four years. Experiments were conducted to answer these questions.

RESULTS

Microarray data were generated in 2009 in three replicates for each of the two MAQC samples with either expired Affymetrix U133A or unexpired U133Plus2 microarrays. These results were compared with data obtained in 2005 on the U133Plus2 microarray. The percentage of overlap between the lists of differentially expressed genes (DEGs) from U133Plus2 microarray data generated in 2009 and in 2005 was 97.44%. While there was some degree of fold change compression in the expired U133A microarrays, the percentage of overlap between the lists of DEGs from the expired and unexpired microarrays was as high as 96.99%. Moreover, the microarray data generated using the expired U133A microarrays in 2009 were highly concordant with microarray and TaqMan® data generated by the MAQC project in 2005.

CONCLUSIONS

Our results demonstrated that microarray data generated using U133A microarrays, which were more than four years past the manufacturer's expiration date, were highly specific and consistent with those from unexpired microarrays in identifying DEGs despite some appreciable fold change compression and decrease in sensitivity. Our data also suggested that the MAQC reference RNA samples, stored at -80°C, were stable over a time frame of at least four years.

Metabolism, genotoxicity, and carcinogenicity of comfrey

Comfrey has been consumed by humans as a vegetable and a tea and used as an herbal medicine for more than 2000 years. Comfrey, however, produces hepatotoxicity in livestock and humans and carcinogenicity in experimental animals. Comfrey contains as many as 14 pyrrolizidine alkaloids (PA), including 7-acetylintermedine, 7-acetyllycopsamine, echimidine, intermedine, lasiocarpine, lycopsamine, myoscorpine, symlandine, symphytine, and symviridine. The mechanisms underlying comfrey-induced genotoxicity and carcinogenicity are still not fully understood. The available evidence suggests that the active metabolites of PA in comfrey interact with DNA in liver endothelial cells and hepatocytes, resulting in DNA damage, mutation induction, and cancer development. Genotoxicities attributed to comfrey and riddelliine (a representative genotoxic PA and a proven rodent mutagen and carcinogen) are discussed in this review. Both of these compounds induced similar profiles of 6,7-dihydro-7-hydroxy-1-hydroxymethyl-5H-pyrrolizine (DHP)-derived DNA adducts and similar mutation spectra. Further, the two agents share common mechanisms of drug metabolism and carcinogenesis. Overall, comfrey is mutagenic in liver, and PA contained in comfrey appear to be responsible for comfrey-induced toxicity and tumor induction.

RDCurve: a nonparametric method to evaluate the stability of ranking procedures

Great concerns have been raised about the reproducibility of gene signatures based on high-throughput techniques such as microarray. Studies analyzing similar samples often report poorly overlapping results, and the p-value usually lacks biological context. We propose a nonparametric ReDiscovery Curve (RDCurve) method, to estimate the frequency of rediscovery of gene signature identified. Given a ranking procedure and a data set with replicated measurements, the RDCurve bootstraps the data set and repeatedly applies the ranking procedure, selects a subset of k important genes, and estimates the probability of rediscovery of the selected subset of genes. We also propose a permutation scheme to estimate the confidence band under the Null hypothesis for the significance of the RDCurve. The method is nonparametric and model-independent. With the RDCurve, we can assess the signal-to-noise ratio of the data, compare the performance of ranking procedures in term of their expected rediscovery rates, and choose the number of genes to be reported.

Localized complement activation in the development of protective immunity against Ostertagia ostertagi infections in cattle

The abomasal nematode Ostertagia ostertagi is a major causal agent contributing to production inefficiencies in the cattle industry in temperate regions of the world. Protective immunity to infections develops very slowly and resistance to reinfection manifests only after prolonged exposure. Mechanisms underlying the development of protective immunity remain largely unexplored. Immune animals, which have significantly reduced worm burdens, were developed after multiple drug-attenuated experimental infections and were compared to a primary infected group and their respective uninfected controls. In this study, transcriptomic analysis identified three signaling pathways significantly impacted during both primary and repeat infections, the complement system, leukocyte extravasation and acute phase responses. Increased mRNA levels of complement components C3, factor B (CFB) and factor I (CFI) in the abomasal mucosa of the infected cattle were confirmed using quantitative PCR while Western blot analysis established the presence of elevated levels of activated C3 proteins in the mucosa. One of the initiators of local complement activation could be related to secretory IgA and IgM because infections significantly up-regulated expression of J chain (IGJ), as well as polymeric Ig receptor (PIGR) and an IgM-specific receptor (FAIM3), suggesting sustained increases in both synthesis and transepithelial transport of IgA and IgM during the infection. The elevated levels of pro-inflammatory cytokines, such as IL-4 and IL-1β, during infection may be involved in gene regulation of complement components. Our results suggest enhanced tissue repair and mucin secretion in immune animals may also contribute to protective immunity. These results are the first evidence that local complement activation may be involved in the development of long-term protective immunity and provide a novel mechanistic insight into resistance against O. ostertagi in cattle.

Pathway modeling of microarray data: a case study of pathway activity changes in the testis following in utero exposure to dibutyl phthalate (DBP)

Pathway activity level analysis, the approach pursued in this study, focuses on all genes that are known to be members of metabolic and signaling pathways as defined by the KEGG database. The pathway activity level analysis entails singular value decomposition (SVD) of the expression data of the genes constituting a given pathway. We explore an extension of the pathway activity methodology for application to time-course microarray data. We show that pathway analysis enhances our ability to detect biologically relevant changes in pathway activity using synthetic data. As a case study, we apply the pathway activity level formulation coupled with significance analysis to microarray data from two different rat testes exposed in utero to Dibutyl Phthalate (DBP). In utero DBP exposure in the rat results in developmental toxicity of a number of male reproductive organs, including the testes. One well-characterized mode of action for DBP and the male reproductive developmental effects is the repression of expression of genes involved in cholesterol transport, steroid biosynthesis and testosterone synthesis that lead to a decreased fetal testicular testosterone. Previous analyses of DBP testes microarray data focused on either individual gene expression changes or changes in the expression of specific genes that are hypothesized, or known, to be important in testicular development and testosterone synthesis. However, a pathway analysis may inform whether there are additional affected pathways that could inform additional modes of action linked to DBP developmental toxicity. We show that Pathway activity analysis may be considered for a more comprehensive analysis of microarray data.

Spatio-temporal regulation of Wnt and retinoic acid signaling by tbx16/spadetail during zebrafish mesoderm differentiation

Background

A complex network of signaling pathways and transcription factors regulates vertebrate mesoderm development. Zebrafish mutants provide a powerful tool for examining the roles of individual genes in such a network. spadetail (spt) is a mutant with a lesion in tbx16, a T-box transcription factor involved in mesoderm development; the mutant phenotype includes disrupted primitive red blood cell formation as well as disrupted somitogenesis. Despite much recent progress, the downstream targets of tbx16 remain incompletely understood. The current study was carried out to test whether any of the five major signaling pathways are regulated by tbx16 during two specific stages of mesoderm development: primitive red blood cell formation in the intermediate mesoderm and somite formation in the tail paraxial mesoderm. This test was performed using Gene Set Enrichment Analysis, which identifies coordinated changes in expression among a priori sets of genes associated with biological features or processes.

Results

Our Gene Set Enrichment Analysis results identify Wnt and retinoic acid signaling as likely downstream targets of tbx16 in the developing zebrafish intermediate mesoderm, the site of primitive red blood cell formation. In addition, such results identify retinoic acid signaling as a downstream target of tbx16 in the developing zebrafish posterior somites. Finally, using candidate gene identification and in situ hybridization, we provide expression domain information for 25 additional genes downstream of tbx16 that are outside of both pathways; 23 were previously unknown downstream targets of tbx16, and seven had previously uncharacterized expression in zebrafish.

Conclusions

Our results suggest that (1) tbx16 regulates Wnt signaling in the developing zebrafish intermediate mesoderm, the site of primitive red blood cell formation, and (2) tbx16 regulates retinoic acid signaling at two distinct embryonic locations and developmental stages, which may imply ongoing spatio-temporal regulation throughout mesoderm development.

Interlaboratory and interplatform comparison of microarray gene expression analysis of HepG2 cells exposed to benzo(a)pyrene

Microarray technology is being used increasingly to study gene expression of biological systems on a large scale. Both interlaboratory and interplatform differences are known to contribute to variability in microarray data. In this study we have investigated data from different platforms and laboratories on the transcriptomic profile of HepG2 cells exposed to benzo(a)pyrene (BaP). RNA samples generated in two different laboratories were analyzed using both Agilent oligonucleotide microarrays and Cancer Research UK (CR-UK) cDNA microarrays. Comparability of the expression profiles was assessed at various levels including correlation and overlap between the data, clustering of the data and affected biological processes. Overlap and correlation occurred, but it was not possible to deduce whether choice of platform or interlaboratory differences contributed more to the data variation. Principal component analysis (PCA) and hierarchical clustering of the expression profiles indicated that the data were most clearly defined by duration of exposure to BaP, suggesting that laboratory and platform variability does not mask the biological effects. Real-time quantitative PCR was used to validate the two array platforms and indicated that false negatives, rather than false positives, are obtained with both systems. All together these results suggest that data from similar biological experiments analyzed on different microarray platforms can be combined to give a more complete transcriptomic profile. Each platform gives a slight variation in the BaP-gene expression response and, although it cannot be stated which is more correct, combining the two data sets is more informative than considering them individually.

LTR: Linear Cross-Platform Integration of Microarray Data

The size and scope of microarray experiments continue to increase. However, datasets generated on different platforms or at different centres contain biases. Improved techniques are needed to remove platform- and batch-specific biases. One experimental control is the replicate hybridization of a subset of samples at each site or on each platform to learn the relationship between the two platforms. To date, no algorithm exists to specifically use this type of control. LTR is a linear-modelling-based algorithm that learns the relationship between different microarray batches from replicate hybridizations. LTR was tested on a new benchmark dataset of 20 samples hybridized to different Affymetrix microarray platforms. Before LTR, the two platforms were significantly different; application of LTR removed this bias. LTR was tested with six separate data pre-processing algorithms, and its effectiveness was independent of the pre-processing algorithm. Sample-size experiments indicate that just three replicate hybridizations can significantly reduce bias. An R library implementing LTR is available.

Changes in gene expression after phencyclidine administration in developing rats: a potential animal model for schizophrenia

Repeated administration of phencyclidine (PCP), an N-methyl-d-aspartate (NMDA) receptor antagonist, during development, may result in neuronal damage that leads to behavioral deficits in adulthood. The present study examined the potential neurotoxic effects of PCP exposure (10mg/kg) in rats on postnatal days (PNDs) 7, 9 and 11 and the possible underlying mechanism(s) for neurotoxicity. Brain tissue was harvested for RNA extraction and morphological assessments. RNA was collected from the frontal cortex for DNA microarray analysis and quantitative RT-PCR. Gene expression profiling was determined using Illumina Rat Ref-12 Expression BeadChips containing 22,226 probes. Based on criteria of a fold-change greater than 1.4 and a P-value less than 0.05, 19 genes including NMDAR1 (N-methyl-d-aspartate receptor) and four pro-apoptotic genes were up-regulated, and 25 genes including four anti-apoptotic genes were down-regulated, in the PCP-treated group. In addition, the schizophrenia-relevant genes, Bdnf (Brain-derived neurotrophic factor) and Bhlhb2 (basic helix-loop-helix domain containing, class B, 2), were significantly different between the PCP and the control groups. Quantitative RT-PCR confirmed the microarray results. Elevated neuronal cell death was further confirmed using Fluoro-Jade C staining. These findings support the hypothesis that neurodegeneration caused by PCP occurs, at least in part, through the up-regulation of NMDA receptors, which makes neurons possessing these receptors more vulnerable to endogenous glutamate. The changes in schizophrenia-relevant genes after repeated PCP exposure during development may provide important information concerning the validation of an animal model for this disorder.

The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models

Gene expression data from microarrays are being applied to predict preclinical and clinical endpoints, but the reliability of these predictions has not been established. In the MAQC-II project, 36 independent teams analyzed six microarray data sets to generate predictive models for classifying a sample with respect to one of 13 endpoints indicative of lung or liver toxicity in rodents, or of breast cancer, multiple myeloma or neuroblastoma in humans. In total, >30,000 models were built using many combinations of analytical methods. The teams generated predictive models without knowing the biological meaning of some of the endpoints and, to mimic clinical reality, tested the models on data that had not been used for training. We found that model performance depended largely on the endpoint and team proficiency and that different approaches generated models of similar performance. The conclusions and recommendations from MAQC-II should be useful for regulatory agencies, study committees and independent investigators that evaluate methods for global gene expression analysis.

Protein phosphatase 1-dependent transcriptional programs for long-term memory and plasticity

Gene transcription is essential for the establishment and the maintenance of long-term memory (LTM) and for long-lasting forms of synaptic plasticity. The molecular mechanisms that control gene transcription in neuronal cells are complex and recruit multiple signaling pathways in the cytoplasm and the nucleus. Protein kinases (PKs) and phosphatases (PPs) are important players in these mechanisms. Protein serine/threonine phosphatase 1 (PP1), in particular, was recently shown to be important for transcription-dependent memory by regulating chromatin remodeling. However, the impact of PP1 on gene transcription in adult neurons remains not fully delineated. Here, we demonstrate that the nuclear pool of PP1 is associated with transcriptional events involving molecular components of signaling cascades acting as positive and negative regulators of memory and brain plasticity. The data show that inhibiting this pool selectively in forebrain neurons improves memory performance, enhances long-term potentiation (LTP), and modulates gene transcription. These findings highlight an important role for PP1 in the regulation of gene transcription in LTM and synaptic plasticity in the adult brain.

Integrative and comparative genomics analysis of early hepatocellular carcinoma differentiated from liver regeneration in young and old

BACKGROUND

Hepatocellular carcinoma (HCC) is the third-leading cause of cancer-related deaths worldwide. It is often diagnosed at an advanced stage, and hence typically has a poor prognosis. To identify distinct molecular mechanisms for early HCC we developed a rat model of liver regeneration post-hepatectomy, as well as liver cells undergoing malignant transformation and compared them to normal liver using a microarray approach. Subsequently, we performed cross-species comparative analysis coupled with copy number alterations (CNA) of independent early human HCC microarray studies to facilitate the identification of critical regulatory modules conserved across species.

RESULTS

We identified 35 signature genes conserved across species, and shared among different types of early human HCCs. Over 70% of signature genes were cancer-related, and more than 50% of the conserved genes were mapped to human genomic CNA regions. Functional annotation revealed genes already implicated in HCC, as well as novel genes which were not previously reported in liver tumors. A subset of differentially expressed genes was validated using quantitative RT-PCR. Concordance was also confirmed for a significant number of genes and pathways in five independent validation microarray datasets. Our results indicated alterations in a number of cancer related pathways, including p53, p38 MAPK, ERK/MAPK, PI3K/AKT, and TGF-beta signaling pathways, and potential critical regulatory role of MYC, ERBB2, HNF4A, and SMAD3 for early HCC transformation.

CONCLUSIONS

The integrative analysis of transcriptional deregulation, genomic CNA and comparative cross species analysis brings new insights into the molecular profile of early hepatoma formation. This approach may lead to robust biomarkers for the detection of early human HCC.

Genomic and metabolomic advances in the identification of disease and adverse event biomarkers

Incomplete knowledge of tissue pathogenesis is hampering the identification of biomarkers for the appropriate therapeutic targets to prevent or inhibit disease processes, and the prediction and diagnosis of injury due to disease and adverse events of drug therapy. The revolution in genomics and metabolomics, combined with advanced bioinformatics and computational methods for mining such large, complex data sets, are beginning to provide critical insights into tissue injury. Such results will move us closer to the promise of personalized medicine.

Shrinkage estimation of effect sizes as an alternative to hypothesis testing followed by estimation in high-dimensional biology: applications to differential gene expression

Research on analyzing microarray data has focused on the problem of identifying differentially expressed genes to the neglect of the problem of how to integrate evidence that a gene is differentially expressed with information on the extent of its differential expression. Consequently, researchers currently prioritize genes for further study either on the basis of volcano plots or, more commonly, according to simple estimates of the fold change after filtering the genes with an arbitrary statistical significance threshold. While the subjective and informal nature of the former practice precludes quantification of its reliability, the latter practice is equivalent to using a hard-threshold estimator of the expression ratio that is not known to perform well in terms of mean-squared error, the sum of estimator variance and squared estimator bias. On the basis of two distinct simulation studies and data from different microarray studies, we systematically compared the performance of several estimators representing both current practice and shrinkage. We find that the threshold-based estimators usually perform worse than the maximum-likelihood estimator (MLE) and they often perform far worse as quantified by estimated mean-squared risk. By contrast, the shrinkage estimators tend to perform as well as or better than the MLE and never much worse than the MLE, as expected from what is known about shrinkage. However, a Bayesian measure of performance based on the prior information that few genes are differentially expressed indicates that hard-threshold estimators perform about as well as the local false discovery rate (FDR), the best of the shrinkage estimators studied. Based on the ability of the latter to leverage information across genes, we conclude that the use of the local-FDR estimator of the fold change instead of informal or threshold-based combinations of statistical tests and non-shrinkage estimators can be expected to substantially improve the reliability of gene prioritization at very little risk of doing so less reliably. Since the proposed replacement of post-selection estimates with shrunken estimates applies as well to other types of high-dimensional data, it could also improve the analysis of SNP data from genome-wide association studies.

Molecular biomarkers: a US FDA effort

Molecular biomarkers are used for various purposes, including disease diagnosis and prognosis, prediction and assessment of treatment response, and safety assessment. There has been a significant increase in the number of US FDA-approved drug labels containing information on molecular biomarkers over the last decade. Almost every pharmaceutical company has been developing molecular biomarker programs, either alone, through partnerships or other ventures. More molecular biomarkers are expected to be identified and validated in drug development, and used to support approval of drug products. This article summarizes the current status of molecular biomarkers used for FDA-approved drug products, and discusses the challenges and future perspectives for the identification and qualification of molecular biomarkers. Specific FDA programs and research projects related to molecular biomarkers are also discussed for supporting regulatory review in the future.

Ginkgo biloba extract induces gene expression changes in xenobiotics metabolism and the Myc-centered network

The use of herbal dietary supplements in the United States is rapidly growing, and it is crucial that the quality and safety of these preparations be ensured. To date, it is still a challenge to determine the mechanisms of toxicity induced by mixtures containing many chemical components, such as herbal dietary supplements. We previously proposed that analyses of the gene expression profiles using microarrays in the livers of rodents treated with herbal dietary supplements is a potentially practical approach for understanding the mechanism of toxicity. In this study, we utilized microarrays to analyze gene expression changes in the livers of male B6C3F1 mice administered Ginkgo biloba leaf extract (GBE) by gavage for 2 years, and to determine pathways and mechanisms associated with GBE treatments. Analysis of 31,802 genes revealed that there were 129, 289, and 2,011 genes significantly changed in the 200, 600, and 2,000 mg/kg treatment groups, respectively, when compared with control animals. Drug metabolizing genes were significantly altered in response to GBE treatments. Pathway and network analyses were applied to investigate the gene relationships, functional clustering, and mechanisms involved in GBE exposure. These analyses indicate alteration in the expression of genes coding for drug metabolizing enzymes, the NRF2-mediated oxidative stress response pathway, and the Myc gene-centered network named "cell cycle, cellular movement, and cancer" were found. These results indicate that Ginkgo biloba-related drug metabolizing enzymes may cause herb-drug interactions and contribute to hepatotoxicity. In addition, the outcomes of pathway and network analysis may be used to elucidate the toxic mechanisms of Ginkgo biloba.

Cross-platform analysis of global microRNA expression technologies

Background

Although analysis of microRNAs (miRNAs) by DNA microarrays is gaining in popularity, these new technologies have not been adequately validated. We examined within and between platform reproducibility of four miRNA array technologies alongside TaqMan PCR arrays.

Results

Two distinct pools of reference materials were selected in order to maximize differences in miRNA content. Filtering for miRNA that yielded signal above background revealed 54 miRNA probes (matched by sequence) across all platforms. Using this probeset as well as all probes that were present on an individual platform, within-platform analyses revealed Spearman correlations of >0.9 for most platforms. Comparing between platforms, rank analysis of the log ratios of the two reference pools also revealed high correlation (range 0.663-0.949). Spearman rank correlation and concordance correlation coefficients for miRNA arrays against TaqMan qRT-PCR arrays were similar for all of the technologies. Platform performances were similar to those of previous cross-platform exercises on mRNA and miRNA microarray technologies.

Conclusions

These data indicate that miRNA microarray platforms generated highly reproducible data and can be recommended for the study of changes in miRNA expression.

Development and validation of a resistance and virulence gene microarray targeting Escherichia coli and Salmonella enterica

A microarray was developed to simultaneously screen Escherichia coli and Salmonella enterica for multiple genetic traits. The final array included 203 60-mer oligonucleotide probes, including 117 for resistance genes, 16 for virulence genes, 25 for replicon markers, and 45 other markers. Validity of the array was tested by assessing inter-laboratory agreement among four collaborating groups using a blinded study design. Internal validation indicated that the assay was reliable (area under the receiver-operator characteristic curve=0.97). Inter-laboratory agreement, however, was poor when estimated using the intraclass correlation coefficient, which ranged from 0.27 (95% confidence interval 0.24, 0.29) to 0.29 (0.23, 0.34). These findings suggest that extensive testing and procedure standardization will be needed before bacterial genotyping arrays can be readily shared between laboratories.

Transcriptome analysis reveals a major impact of JAK protein tyrosine kinase 2 (Tyk2) on the expression of interferon-responsive and metabolic genes

Background

Tyrosine kinase 2 (Tyk2), a central component of Janus kinase/signal transducer and activator of transcription (JAK/STAT) signaling, has major effects on innate immunity and inflammation. Mice lacking Tyk2 are resistant to endotoxin shock induced by lipopolysaccharide (LPS), and Tyk2 deficient macrophages fail to efficiently induce interferon α/β after LPS treatment. However, how Tyk2 globally regulates transcription of downstream target genes remains unknown. Here we examine the regulatory role of Tyk2 in basal and inflammatory transcription by comparing gene expression profiles of peritoneal macrophages from Tyk2 mutant and wildtype control mice that were either kept untreated or exposed to LPS for six hours.

Results

Untreated Tyk2-deficient macrophages exhibited reduced expression of immune response genes relative to wildtype, in particular those that contain interferon response elements (IRF/ISRE), whereas metabolic genes showed higher expression. Upon LPS challenge, IFN-inducible genes (including those with an IRF/ISRE transcription factor binding-site) were strongly upregulated in both Tyk2 mutant and wildtype cells and reached similar expression levels. In contrast, metabolic gene expression was strongly decreased in wildtype cells upon LPS treatment, while in Tyk2 mutant cells the expression of these genes remained relatively unchanged, which exaggerated differences already present at the basal level. We also identified several 5'UR transcription factor binding-sites and 3'UTR regulatory elements that were differentially induced between Tyk2 deficient and wildtype macrophages and that have not previously been implicated in immunity.

Conclusions

Although Tyk2 is essential for the full LPS response, its function is mainly required for baseline expression but not LPS-induced upregulation of IFN-inducible genes. Moreover, Tyk2 function is critical for the downregulation of metabolic genes upon immune challenge, in particular genes involved in lipid metabolism. Together, our findings suggest an important regulatory role for Tyk2 in modulating the relationship between immunity and metabolism.

Transcriptional profiling of the acute pulmonary inflammatory response induced by LPS: role of neutrophils

Background

Lung cancer often develops in association with chronic pulmonary inflammatory diseases with an influx of neutrophils. More detailed information on inflammatory pathways and the role of neutrophils herein is a prerequisite for understanding the mechanism of inflammation associated cancer.

Methods

In the present study, we used microarrays in order to obtain a global view of the transcriptional responses of the lung to LPS in mice, which mimics an acute lung inflammation. To investigate the influence of neutrophils in this process, we depleted mice from circulating neutrophils by treatment with anti-PMN antibodies prior to LPS exposure.

Results

A total of 514 genes was greater than 1.5-fold differentially expressed in the LPS induced lung inflammation model. 394 of the 514 were up regulated genes mostly involved in cell cycle and immune/inflammation related processes, such as cytokine/chemokine activity and signalling. Down regulated genes represented nonimmune processes, such as development, metabolism and transport. Notably, the number of genes and pathways that were differentially expressed, was reduced when animals were depleted from circulating neutrophils, confirming the central role of neutrophils in the inflammatory response. Furthermore, there was a significant correlation between the differentially expressed gene list and the promutagenic DNA lesion M₁dG, suggesting that it is the extent of the immune response which drives genetic instability in the inflamed lung. Several genes that were specifically regulated by the presence of activated neutrophils could be identified and these were mostly involved in interferon signalling, oxidative stress response and cell cycle progression. The latter possibly refers to a higher rate of cell turnover in the inflamed lung with neutrophils, suggesting that the neutrophil influx is associated with a higher risk for the accumulation and fixation of mutations.

Conclusion

Gene expression profiling identified specific genes and pathways that are related to neutrophilic inflammation and could be associated to cancer development and indicate an active role of neutrophils in mediating the LPS induced inflammatory response in the mouse lung.

Abnormal gene expression in cerebellum of Npc1-/- mice during postnatal development

Niemann-Pick Type C (NPC) disease is an autosomal recessive neurodegenerative disorder with abnormal lipid storage as the major cellular pathologic hallmark. Genetic analyses have identified mutations in NPC1 gene in the great majority of cases, while mutations in NPC2 account for the remainders. Yet little is known regarding the cellular mechanisms responsible for NPC pathogenesis, especially for neurodegeneration, which is the usual cause of death. To identify critical steps that could account for the pathological manifestations of the disease in one of the most affected brain structures, we performed global gene expression analysis in the cerebellum from 3-week old Npc1+/+ and Npc1-/- mice with two different microarray platforms (Agilent and Illumina). Differentially expressed genes identified by both microarray platforms were then subjected to KEGG pathway analysis. Expression of genes in six pathways was significantly altered in Npc1-/- mice; functionally, these signaling pathways belong to the following three categories: (1) steroid and terpenoid biosynthesis, (2) immune response, and (3) cell adhesion/motility. In addition, the expression of several proteins involved in lipid transport was significantly altered in Npc1-/- mice. Our results provide novel molecular insight regarding the mechanisms of pathogenesis in NPC disease and reveal potential new therapeutic targets.

Validation of differential gene expression algorithms: application comparing fold-change estimation to hypothesis testing

BACKGROUND

Sustained research on the problem of determining which genes are differentially expressed on the basis of microarray data has yielded a plethora of statistical algorithms, each justified by theory, simulation, or ad hoc validation and yet differing in practical results from equally justified algorithms. Recently, a concordance method that measures agreement among gene lists have been introduced to assess various aspects of differential gene expression detection. This method has the advantage of basing its assessment solely on the results of real data analyses, but as it requires examining gene lists of given sizes, it may be unstable.

RESULTS

Two methodologies for assessing predictive error are described: a cross-validation method and a posterior predictive method. As a nonparametric method of estimating prediction error from observed expression levels, cross validation provides an empirical approach to assessing algorithms for detecting differential gene expression that is fully justified for large numbers of biological replicates. Because it leverages the knowledge that only a small portion of genes are differentially expressed, the posterior predictive method is expected to provide more reliable estimates of algorithm performance, allaying concerns about limited biological replication. In practice, the posterior predictive method can assess when its approximations are valid and when they are inaccurate. Under conditions in which its approximations are valid, it corroborates the results of cross validation. Both comparison methodologies are applicable to both single-channel and dual-channel microarrays. For the data sets considered, estimating prediction error by cross validation demonstrates that empirical Bayes methods based on hierarchical models tend to outperform algorithms based on selecting genes by their fold changes or by non-hierarchical model-selection criteria. (The latter two approaches have comparable performance.) The posterior predictive assessment corroborates these findings.

CONCLUSIONS

Algorithms for detecting differential gene expression may be compared by estimating each algorithm's error in predicting expression ratios, whether such ratios are defined across microarray channels or between two independent groups.According to two distinct estimators of prediction error, algorithms using hierarchical models outperform the other algorithms of the study. The fact that fold-change shrinkage performed as well as conventional model selection criteria calls for investigating algorithms that combine the strengths of significance testing and fold-change estimation.

Gene expression profiling in the developing rat brain exposed to ketamine

Ketamine, a non-competitive N-methyl-d-aspartate (NMDA) receptor antagonist, is associated with accelerated neuronal apoptosis in the developing rodent brain. In this study, postnatal day (PND) 7 rats were treated with 20 mg/kg ketamine or saline in six successive doses (s.c.) at 2-h intervals. Brain frontal cortical areas were collected 6 h after the last dose and RNA isolated and hybridized to Illumina Rat Ref-12 Expression BeadChips containing 22,226 probes. Many of the differentially expressed genes were associated with cell death or differentiation and receptor activity. Ingenuity Pathway Analysis software identified perturbations in NMDA-type glutamate, GABA and dopamine receptor signaling. Quantitative polymerase chain reaction (Q-PCR) confirmed that NMDA receptor subunits were significantly up-regulated. Up-regulation of NMDA receptor mRNA signaling was further confirmed by in situ hybridization. These observations support our working hypothesis that prolonged ketamine exposure produces up-regulation of NMDA receptors and subsequent over-stimulation of the glutamatergic system by endogenous glutamate, triggering enhanced apoptosis in developing neurons.

DNA microarrays are predictive of cancer prognosis: a re-evaluation

PURPOSE

The reliability of microarray-based cancer prognosis is questioned by Michiels et al. They reanalyzed seven studies published in the prominent journals as successful stories of microarray-based cancer prognosis and concluded that the originally reported assessments are over optimistic. We set to investigate the reality of microarrays for predicting cancer prognosis by using the same data sets with commonly accepted data analysis approaches.

EXPERIMENT DESIGN

Michiels et al.'s analysis protocol used a correlation-based feature selection method, split sample validation, and a nearest-centroid rule classifier. We examined their results through systematically replacing their analysis approaches with other commonly used methods as a parameter study. In addition, we applied a widely accepted permutation test in conjunction with 5-fold cross-validation to verify Michiels et al.'s findings.

RESULTS

The stability of signature genes is likely obtained when a fold change-based feature selection method is applied. When cross-validation procedures are used to replace Michiels et al.'s split sample validation, only one of the seven studies yielded uninformative classifiers. The permutation test reveals that the confidence interval based on the split sample used in the Michiels et al.'s review is not a rigorous and robust approach to assess the validity of a classifier.

CONCLUSIONS

We concluded that the use of DNA microarrays for cancer prognosis can be demonstrated. We also stressed that caution should be exercised when a general conclusion is withdrawn based on a single statistical practice without alternative validation, which can leave a false impression and pessimistic perspective for emerging biomarker methodologies to advance cancer research.

Application of microarray-based analysis of gene expression in the field of toxicogenomics

The field of toxicogenomics, which is becoming an important sub-discipline of toxicology, resulted from the natural convergence of the field of conventional toxicological research and the emergent field of functional genomics. One technology that has played a significant role in the field of toxicogenomics (in addition to many others) is the gene expression microarray. In this chapter, the authors provide an example of the application of gene expression microarrays to the field of toxicogenomics by detailing the strategy that was used for obtaining, analyzing, and interpreting gene expression data generated from RNA isolated from the liver of toxicant-exposed rats.

Probing differentially expressed genes against a microarray database for in silico suppressor/enhancer and inhibitor/activator screens

High-density oligonucleotide arrays are widely used for analysis of gene expression on a genomic scale, but the generated data remain largely inaccessible for comparative analysis purposes. Similarity searches in databases with differentially expressed gene (DEG) lists may be used to assign potential functions to new genes and to identify potential chemical inhibitors/activators and genetic suppressors/enhancers. Although this is a very promising concept, it requires the compatibility and validity of the DEG lists to be significantly improved. Using Arabidopsis and human datasets, we have developed guidelines for the performance of similarity searches against databases that collect microarray data. We found that, in comparison with many other methods, a rank-product analysis achieves a higher degree of inter- and intra-laboratory consistency of DEG lists, and is advantageous for assessing similarities and differences between them. To support this concept, we developed a tool called MASTA (microarray overlap search tool and analysis), and re-analyzed over 600 Arabidopsis microarray expression datasets. This revealed that large-scale searches produce reliable intersections between DEG lists that prove to be useful for genetic analysis, thus aiding in the characterization of cellular and molecular mechanisms. We show that this approach can be used to discover unexpected connections and to illuminate unanticipated interactions between individual genes.

Introduction to the development and validation of predictive biomarker models from high-throughput data sets

High-throughput technologies can routinely assay biological or clinical samples and produce wide data sets where each sample is associated with tens of thousands of measurements. Such data sets can be mined to discover biomarkers and develop statistical models capable of predicting an endpoint of interest from data measured in the samples. The field of biomarker model development combines methods from statistics and machine learning to develop and evaluate predictive biomarker models. In this chapter, we discuss the computational steps involved in the development of biomarker models designed to predict information about individual samples and review approaches often used to implement each step. A practical example of biomarker model development in a large gene expression data set is presented. This example leverages BDVal, a suite of biomarker model development programs developed as an open-source project (see http://bdval.org /).

Global gene expression analysis reveals differences in cellular responses to hydroxyl- and superoxide anion radical-induced oxidative stress in caco-2 cells

Reactive oxygen species-induced oxidative stress in the colon is involved in inflammatory bowel diseases and suggested to be associated with colorectal cancer risk. However, our insight in molecular responses to different oxygen radicals is still fragmentary. Therefore, we studied global gene expression by an extensive time series (0.08, 0.25, 0.5, 1, 2, 4, 8, 16, or 24 h) analyses in human colon cancer (caco-2) cells after exposure to H(2)O(2) or the superoxide anion donor menadione. Differences in gene expression were investigated by hybridization on two-color microarrays against nonexposed time-matched control cells. Next to gene expression, correlations with related phenotypic markers (8-oxodG levels and cell cycle arrest) were investigated. Gene expression analysis resulted in 1404 differentially expressed genes upon H(2)O(2) challenge and 979 genes after menadione treatment. Further analysis of gene expression data revealed how these oxidant responses can be discriminated. Time-dependent coregulated genes immediately showed a pulse-like response to H(2)O(2), while the menadione-induced expression is not restored over 24 h. Pathway analyses demonstrated that H(2)O(2) immediately influences pathways involved in the immune function, while menadione constantly regulated cell cycle-related pathways Altogether, this study offers a novel and detailed insight in the similarities and differences of the time-dependent oxidative stress responses induced by the oxidants H(2)O(2) and menadione and show that these can be discriminated regarding their modulation of particular colon carcinogenesis-related mechanisms.

A novel approach to analyze gene expression data demonstrates that the DeltaF508 mutation in CFTR downregulates the antigen presentation pathway

Gene array studies comparing cystic fibrosis (CF) and non-CF genotypes should reveal factors that explain variability in CF lung disease progression, yielding insights that lead to improved CF care. To date, studies have reached conflicting conclusions, perhaps due to experimental differences and divergent statistical approaches. This review aims: 1) to summarize the findings of four recent gene studies comparing CF and non-CF genotypes, and 2) to reanalyze original data using a recently developed statistical approach, with the aim of identifying genes and paths consistently regulated by the CF genotype. We identified four studies evaluating the effect of the DeltaF508-CFTR mutation on human airway epithelial cell gene expression, restricting our investigation to human airway epithelial cell studies whose data were accessible in NCBI's Gene Expression Omnibus or the European Bioinformatic Institute's ArrayExpress. Gene expression patterns showed consistent repression of MHC class I antigen presentation genes in CF human airway epithelia, suggesting a novel mechanistic explanation for poor clearance of viral and bacterial infections by CF patients. We also examined proinflammatory and NF-kappaB genes, whose induction is widely accepted as a hallmark of the CF genotype, but found little evidence of induction, consistent with a recent review (Machen TE, Am J Physiol Cell Physiol 291: C218-C230, 2006.). In conclusion, our analysis suggests that the CF genotype may impair immune function in airway epithelial cells but may not increase inflammation. Additional studies are required to determine whether MHC class I gene repression in CF reduces antigen presentation at the protein level and whether repression impairs immune function.

Analysis of genome-wide association data by large-scale Bayesian logistic regression

Single-locus analysis is often used to analyze genome-wide association (GWA) data, but such analysis is subject to severe multiple comparisons adjustment. Multivariate logistic regression is proposed to fit a multi-locus model for case-control data. However, when the sample size is much smaller than the number of single-nucleotide polymorphisms (SNPs) or when correlation among SNPs is high, traditional multivariate logistic regression breaks down. To accommodate the scale of data from a GWA while controlling for collinearity and overfitting in a high dimensional predictor space, we propose a variable selection procedure using Bayesian logistic regression. We explored a connection between Bayesian regression with certain priors and L1 and L2 penalized logistic regression. After analyzing large number of SNPs simultaneously in a Bayesian regression, we selected important SNPs for further consideration. With much fewer SNPs of interest, problems of multiple comparisons and collinearity are less severe. We conducted simulation studies to examine probability of correctly selecting disease contributing SNPs and applied developed methods to analyze Genetic Analysis Workshop 16 North American Rheumatoid Arthritis Consortium data.