Microarray-based Gene Expression Data Research Articles

In silico identification of an epithelial core signature in human tumors.

10628 Background: Gene expression analysis is performed on grossly selected specimens often without any microscopic analysis of tumor content. In studies where histological analyses have been performed, cases having 80% or more tumor content are used for microarray analysis. The variability in amount of epithelial and stromal cells may generate to misleading differential expression analysis and selection for wrong targets for therapeutics. It is also often unclear, whether the genes identified are stromal or epithelial in origin. The goal of this study was to identify genes that define core epithelial phenotype; these genes could provide means of normalization of expression data. Methods: The CABIG GSK microarray (HG-U133_plus_2) data consisting of 950 cell lines from carcinoma (n=562), non-carcinoma (n=385) and normal tissue (n=3) was analyzed to identify epithelial specific genes. 10 carcinomas each from 11 sites (n=110) and an equal number of non-carcinomas were randomly selected. In silico analyses were performed by 1) identifying genes differentially expressed between carcinoma and non-carcinoma samples using a one way ANOVA; 2) identifying gene signature associated with carcinoma using Predictive Analysis of Microarrays (PAM) and 3) a weighted gene coexpression network analysis (WGCNA) was performed to identify co-expression modules. A similar analysis was also performed on tissue samples (E-GEOD-12360) from carcinomas and non-carcinomas. Venn-diagram was generated to identify intersecting set. Results: Comparison of the carcinoma and non-carcinoma samples using ANOVA identified 1455 differential expressed gene probes in cell lines and 540 gene probes in tissues (FDR=1E-10). The cell lines analysis identified 5 modules and a 65-gene signature (43 core and 22 accessory set) that was specific for epithelial cells. In the tissue analysis a 188-gene signature was similarly identified. Cross-comparison identified a smaller 31 gene intersecting set; this was not associated with loss of discriminatory power. Conclusions: A 31 geneset which can be used to determine the epithelial content of heterogeneous tumors, was identified. This study has the potential to significantly impact the use of microarray based gene expression data.

Identification of sepsis subtypes in critically ill adults using gene expression profiling

IntroductionSepsis is a syndromic illness that has traditionally been defined by a set of broad, highly sensitive clinical parameters. As a result, numerous distinct pathophysiologic states may meet diagnostic criteria for sepsis, leading to syndrome heterogeneity. The existence of biologically distinct sepsis subtypes may in part explain the lack of actionable evidence from clinical trials of sepsis therapies. We used microarray-based gene expression data from adult patients with sepsis in order to identify molecularly distinct sepsis subtypes.MethodsWe used partitioning around medoids (PAM) and hierarchical clustering of gene expression profiles from neutrophils taken from a cohort of septic patients in order to identify distinct subtypes. Using the medoids learned from this cohort, we then clustered a second independent cohort of septic patients, and used the resulting class labels to evaluate differences in clinical parameters, as well as the expression of relevant pharmacogenes.ResultsWe identified two sepsis subtypes based on gene expression patterns. Subtype 1 was characterized by increased expression of genes involved in inflammatory and Toll receptor mediated signaling pathways, as well as a higher prevalence of severe sepsis. There were differences between subtypes in the expression of pharmacogenes related to hydrocortisone, vasopressin, norepinephrine, and drotrecogin alpha.ConclusionsSepsis subtypes can be identified based on different gene expression patterns. These patterns may generate hypotheses about the underlying pathophysiology of sepsis and suggest new ways of classifying septic patients both in clinical practice, and in the design of clinical trials.

Kinome expression profiling and prognosis of basal breast cancers

BackgroundBasal breast cancers (BCs) represent ~15% of BCs. Although overall poor, prognosis is heterogeneous. Identification of good- versus poor-prognosis patients is difficult or impossible using the standard histoclinical features and the recently defined prognostic gene expression signatures (GES). Kinases are often activated or overexpressed in cancers, and constitute targets for successful therapies. We sought to define a prognostic model of basal BCs based on kinome expression profiling.MethodsDNA microarray-based gene expression and histoclinical data of 2515 early BCs from thirteen datasets were collected. We searched for a kinome-based GES associated with disease-free survival (DFS) in basal BCs of the learning set using a metagene-based approach. The signature was then tested in basal tumors of the independent validation set.ResultsA total of 591 samples were basal. We identified a 28-kinase metagene associated with DFS in the learning set (N = 73). This metagene was associated with immune response and particularly cytotoxic T-cell response. On multivariate analysis, a metagene-based predictor outperformed the classical prognostic factors, both in the learning and the validation (N = 518) sets, independently of the lymphocyte infiltrate. In the validation set, patients whose tumors overexpressed the metagene had a 78% 5-year DFS versus 54% for other patients (p = 1.62E-4, log-rank test).ConclusionsBased on kinome expression, we identified a predictor that separated basal BCs into two subgroups of different prognosis. Tumors associated with higher activation of cytotoxic tumor-infiltrative lymphocytes harbored a better prognosis. Such classification should help tailor the treatment and develop new therapies based on immune response manipulation.

The other triple-negative breast cancer: Immunohistochemical and clinicopathologic characterization of the Claudin-low subtype.

1129 Background: The triple-negative (TN) immunohistochemical (IHC) definition for basal-like breast cancer lacks specificity. This aggressive subtype is best identified with the addition of EGFR and CK5/6 biomarkers; however, the subgroup of TN breast cancers that lack expression of these biomarkers (5NP) is mostly uncharacterized. Recently, an additional intrinsic subtype, claudin-low, was identified using microarray-based gene expression data. These tumors are characterized by the low expression of ERBB2, ESR1, and cell-cell adhesion proteins such as claudins 3/4/7 and E-cadherin. Methods: Microarray gene expression and IHC data were collected from two independent cohorts: Washington University (WU, n=114) and University of North Carolina (UNC, n=98). Tumors were assigned as claudin-low or not using a genomic predictor. An IHC surrogate for the claudin-low subtype was developed, and applied to a tissue microarray (TMA) from the University of British Columbia (UBC, n=3473) linked with extensive clinicopathological and outcome data. Results: Against a gene expression gold standard, an IHC combination of ER(-), Her2(-), and E-cadherin(-) best identified claudin-low tumors (PPV=24% WU, 67% UNC; NPV=94% WU, 87% UNC). This IHC panel identified 226 claudin-low tumors (6.5%) from the UBC TMA. Compared to the basal-like tumors, claudin-low tumors had a similar distribution for patient age, tumor size and lymph node status. However, the claudin-low tumors were less often grade 3 (88% vs. 80%, p = 0.036) or high expressors of Ki67 (61% vs. 82%, p <0.001). Claudin-low subgroup had superior breast cancer specific survival when compared to the basal-like (10-yr BCSS 69% vs. 60%) with a multivariate HR of 0.69 (p = 0.030), adjusted for clinicopathological variables and adjuvant treatment. Conclusions: The ER(-)/Her2(-)/E-cadherin(-) IHC definition identified the claudin-low subtype of breast cancer, although with a PPV insufficient for clinical use. The addition of this definition to the standard IHC panel for intrinsic breast cancer subtypes (ER, PR, Her2, Ki67, EGFR and CK5/6) isolates a subgroup of TN tumors that are clinically distinct from basal-like breast cancers.

Disease and phenotype gene set analysis of disease-based gene expression in mouse and human

The genetic contributions to common disease and complex disease phenotypes are pleiotropic, multifactorial, and combinatorial. Gene set analysis is a computational approach used in the analysis of microarray data to rapidly query gene combinations and multifactorial processes. Here we use novel gene sets based on population-based human genetic associations in common human disease or experimental genetic mouse models to analyze disease-related microarray studies. We developed a web-based analysis tool that uses these novel disease- and phenotype-related gene sets to analyze microarray-based gene expression data. These gene sets show disease and phenotype specificity in a species-specific and cross-species fashion. In this way, we integrate population-based common human disease genetics, mouse genetically determined phenotypes, and disease or phenotype structured ontologies, with gene expression studies relevant to human disease. This may aid in the translation of large-scale high-throughput datasets into the context of clinically relevant disease phenotypes.

Coexpression network analysis of neural tissue reveals perturbations in developmental processes in schizophrenia

We performed integrated gene coexpression network analysis on two large microarray-based brain gene expression data sets generated from the prefrontal cortex obtained post-mortem from 101 subjects, 47 subjects with schizophrenia and 54 normal control subjects, ranging in age from 19 to 81 years. Twenty-eight modules of coexpressed genes with functional interpretations were detected in both normal subjects and those with schizophrenia. Significant overlap of "case" and "control" module composition was observed, indicating that extensive differences in underlying molecular connectivity are not likely driving pathology in schizophrenia. Modules of coexpressed genes were characterized according to disease association, cell type specificity, and the effects of aging. We find that genes with altered expression in schizophrenia clustered into distinct coexpression networks and that these were associated primarily with neurons. We further identified a robust effect of age on gene expression modules that differentiates normal subjects from those with schizophrenia. In particular, we report that normal age-related decreases in genes related to central nervous system developmental processes, including neurite outgrowth, neuronal differentiation, and dopamine-related cellular signaling, do not occur in subjects with schizophrenia during the aging process. Extrapolating these findings to earlier stages of development supports the concept that schizophrenia pathogenesis begins early in life and is associated with a failure of normal decreases in developmental-related gene expression. These findings provide a novel mechanism for the "developmental" hypothesis of schizophrenia on a molecular level.

Co-expression networks: graph properties and topological comparisons

Abstract Motivation: Microarray-based gene expression data have been generated widely to study different biological processes and systems. Gene co-expression networks are often used to extract information about groups of genes that are ‘functionally’ related or co-regulated. However, the structural properties of such co-expression networks have not been rigorously studied and fully compared with known biological networks. In this article, we aim at investigating the structural properties of co-expression networks inferred for the species Saccharomyces Cerevisiae and comparing them with the topological properties of the known, well-established transcriptional network, MIPS physical network and protein–protein interaction (PPI) network of yeast. Results: These topological comparisons indicate that co-expression networks are not distinctly related with either the PPI or the MIPS physical interaction networks, showing important structural differences between them. When focusing on a more literal comparison, vertex by vertex and edge by edge, the conclusion is the same: the fact that two genes exhibit a high gene expression correlation degree does not seem to obviously correlate with the existence of a physical binding between the proteins produced by these genes or the existence of a MIPS physical interaction between the genes. The comparison of the yeast regulatory network with inferred yeast co-expression networks would suggest, however, that they could somehow be related. Conclusions: We conclude that the gene expression-based co-expression networks reflect more on the gene regulatory networks but less on the PPI or MIPS physical interaction networks. Contact: hongzhe@mail.med.upenn.edu Supplementary information: Supplementary data are available at Bioinformatics online.

마이크로어레이 데이터 공유 시스템

최근, 마이크로어레이 실험 데이터의 품질과 재생산성에 대한 신뢰도가 증가하고 있어 마이크로어레이 데이터의 공유 및 활용에 대한 요구가 급속히 증가하고 있다. 그러나 공개되어 있는 국내, 외 마이크로어레이 데이터는 실험 방식, 플랫폼 등에 따라 서로 다른 데이터 항목과 포맷을 가지므로 데이터의 실제적 접근 및 활용이 어려운 상황이다. 본 논문에서는 실험 플랫폼, 데이터 포맷, 정규화 기법, 분석 방식 등이 서로 다른 기존의 마이크로어레이 데이터를 효율적으로 검색, 공유, 통합할 수 있는 마이크로어레이 데이터 공유 시스템을 제안한다. 제안된 시스템은 웹 서비스 기반 기술을 이용하여 분산된 마이크로어레이 데이터를 통합하며, 각 사이트의 사용자는 UDDI를 통하여 검색한 데이터를 표준 MGED 기반의 공통 데이터 구조로 자동 변환하여 다운 받을 수 있다. 정의된 공통 데이터 구조는 IDF,ADF,SDRF,EDF로 구성되어 다양한 구조의 마이크로어레이를 통합할 수 있는 템플릿 역할을 수행하며, MAGE-ML, MAGE-TAB, XML Schema 문서로 저장할 수 있다. 또한 제안된 시스템의 자동 데이터 제출기, 파일 관리자 등은 마이크로어레이 데이터 공유를 위한 다양한 부가 기능을 제공한다. Improved reliability of microarray data and its reproducibility lead to recent increment in demand of data sharing and utilization among laboratories, but house-keeping and publicly opened microarray experimental data can hardly be accessed and utilized since they are in heterogeneous formats according to the various experimental methods and microarray platforms. In this paper, we propose a microarray sharing method which can easily retrieve and integrate microarray data from different experiment platforms, data formats, normalization methods, and analysis methods. Our system is based on web-service technology. The biologists of each site are able to search UDDI(Universal Description, Discovery, and Integration) registry, and download microarray data with common data structure of standard format recommended by MGED(Microarray Gene Expression Databases) society. The common data structure defined in this paper consists of IDF(Investigation Design Format), ADF(Array Design Format), SDRF(Sample and Relationship Format), and EDF(Expression Data Format). These components play role as templates to integrate microarray data with various structure and can be stored in standard formats such as MAGE-ML, MAGE-TAB, and XML Schema. In addition, our system provides advanced tools of automatic microarray data submitter and file manager to manipulate local microarray data efficiently.

The Differential Gene Expression Pattern ofMycobacterium tuberculosisin Response to Capreomycin and PA-824 versus First-Line TB Drugs Reveals Stress- and PE/PPE-Related Drug Targets

Tuberculosis is a leading infectious disease causing millions of deaths each year. How to eradicate mycobacterial persistence has become a central research focus for developing next-generation TB drugs. Yet, the knowledge in this area is fundamentally limited and only a few drugs, notably capreomycin and PA-824, have been shown to be active against non-replicating persistent TB bacilli. In this study, we performed a new bioinformatics analysis on microarray-based gene expression data obtained from the public domain to explore genes that were differentially induced by drugs between the group of capreomycin and PA-824 and the group of mainly the first-line TB drugs. Our study has identified 42 genes specifically induced by capreomycin and PA-824. Many of these genes are related to stress responses. In terms of the distribution of identified genes in a specific category relative to the whole genome, only the categories of PE/PPE and conserved hypotheticals have statistical significance. Six among the 42 genes identified in this study are on the list of the top 100 persistence targets selected by the TB Structural Genomics Consortium. Further biological elucidation of their roles in mycobacterial persistence is warranted.

Regulation of CD9 Expression by Runx1 in Acute Myeloid Leukemia.

CD9 is a tetraspanin molecule that is expressed on precursor B cells, megakaryocytes, and certain acute myeloid leukemias. CD9 has been show to regulate important integrin-mediated functions such as adhesion and motility and has been found to be an independent adverse risk factor in the prognosis of AML. CD9 is also typically expressed in precursor B cell ALL. However, in pediatric precursor B cell ALL harboring the t(12;21) translocation, CD9 is often absent or its expression is very low. This translocation results in the presence of a fusion protein between TEL and Runx1, which is believed to exert a dominant negative effect on normal Runx1 function. This raises the possibility that CD9 may be a Runx1 transcriptional target. In order to study the role of Runx1 in the regulation of CD9 expression, we examined the role of Runx1 in the regulation of CD9 expression in K562 erythroleukemia cells. CD9 expression is induced by PKC signaling which can be blocked by the ERK signaling inhibitor U0126. The proximal CD9 promoter has previously been shown to have PKC-inducible activity. This core promoter has a canonical Runx1 binding site in it. Using chromatin immunoprecipitation assay, we show that Runx1 is recruited to this region of the CD9 promoter in K562 cells by PKC activation. In order to further evaluate the role of Runx1 in CD9 expression, we expressed Runx1-KRAB-ER in K562 cells. This artificial dominant-negative form of Runx1 utilizes the Runx1 DNA binding domain, a KRAB repressor domain, and a modified human estrogen receptor. In K562 cells expressing Runx1-KRAB-ER, CD9 induction by PKC activation was significantly reduced in the presence of 4-hydroxy-tamoxifen, which releases ER-fused proteins sequestered with heat shock proteins. This data demonstrates that inhibition of normal Runx1 activity interferes with induction of CD9 expression.Similar to precursor B cell ALL with the t(12;21), AML with the t(8;21) has Runx1 fused to ETO and the Runx1/ETO fusion protein has also been reported to repress normal Runx1 function. In order to evaluate CD9 expression in different cytogenetically defined subgroups of AML, we interrogated two large, publicly available AML microarray gene expression data bases (referenced below). We evaluated the expression of CD19, known to be upregulated in t(8;21) AMLs, and IL-3, a known target of Runx1 transcription, as well as CD9. As compared with expression in t(15;17), inv(16;16), and those with normal karyotypes, expression of IL-3 and CD9 were significantly reduced in t(8;21) AML as compared with the other subgroups. In contrast, CD19 expression was greater in t(8;21) AML than the other subgroups. These findings show that CD9 expression is specifically reduced in t(8;21) AMLs as compared to the other cytogenetically-defined subgroups. These findings suggest that the Runx1/ETO protein may be actively repressing CD9 expression in t(8;21) AML. In order to test this hypothesis, we examined the expression of CD9 in Kasumi-1 cells which harbor this translocation. We could not detect CD9 expression at the mRNA or protein levels, nor could PKC activation induce its expression. However, if Kasumi-1 cells were treated with valproic acid, which has been previously shown to disrupt the Runx1/ETO-HDAC1 repressor complex, then CD9 induction by PKC activation is restored. Taken together, these data provide strong evidence that Runx1 regulates CD9 expression and that the absence of CD9 expression in acute leukemias harboring Runx1 fusion proteins likely relates to their repressive effects on normal Runx1 regulation of CD9 expression.

Functional genomics of human pre-implantation development

Early mammalian embryogenesis is currently the focus of intense interest because of the potential of inner cell mass-derived embryonic stem cell lines in new therapeutic strategies. As such, creating molecular profiles of gene expression during pre-implantation development will provide a framework for understanding the biological properties of these cells and also establish a tool set for subsequent functional studies. However, a major obstacle impeding progress in this area are moral issues regarding their use, the scarcity of these cells and the ability to successfully isolate and amplify enough mRNA from the minute amounts of total RNA present in these cells. The elucidation, unravelling and understanding the molecular basis of transcriptional control during pre-implantation development is of utmost importance if we are to diagnose, intervene, eliminate or reduce abnormalities associated with growth, disease and infertility by applying assisted reproduction. Importantly, these studies should enhance our knowledge of basic reproductive biology and its application to regenerative medicine. This review describes the application of in silico-based approaches, in order to obtain maximal information from published microarray-based gene expression data. For an illustration of this, we used gene expression data related to unfertilized oocytes and blastocysts to gain insights into genes and related signalling pathways (e.g. MAPK, PI3K, WNT, TGF-beta) involved in the switch from maternal to embryonic control of gene transcription during human pre-implantation development.

Toxicogenomics analysis correlates attenuation of Nrf2 function with renal toxicity in the rat

In order to gain understanding of mechanisms of chemically-induced renal toxicity, we first curated 4 key biological pathways focused on kidney function and injury from a thorough review of the literature. Genes in these 4 pathways were then used to systematically query and analyze kidney data in DrugMatrix,® a large-scale contextual microarray gene expression database of drug-treated rat tissues. Data mining revealed a tightly-clustered set of experiments, including treatments with compounds of well documented renal toxicity such as 4, 4′-methylenedianiline, 6-mercaptopurine, cerivastatin, rifampin and thalidomide. Pathway analysis revealed that genes involved in TGF-β signaling, inflammation and oxidative stress responses are consistently upregulated by these compounds. However, heme-oxygenase-1 (HO-1), a key cytoprotective mediator of oxidative stress response in kidney, was not up-regulated, while many transcription factors that activate HO-1 transcription were consistently up-regulated. Further analysis revealed that this group of compounds induced the expression of an oxidative stress response suppressor, Bach1, and inhibited the function of Nrf2 which mediates the amplification of oxidative stress response in mammalian cells including the gene expression of HO-1. Thus, Bach1 induction may be an important mechanism for compound-induced renal dysfunction.

A database study that identifies genes whose expression correlates, negatively or positively, with 5-year survival of cancer patients

A published microarray gene expression database containing data on 174 tumor samples from ten tissues was mined, enabling the identification of classes of genes whose expression correlates significantly with the intractability, or tractability, to therapy of tumors derived from such tissues. As a measure of tractability, the 5-year survival of patients presenting with distant (metastatic) tumors was used. Genes that encode proteins related to cell adhesion, and enzymes involved in metabolic oxidation or reduction, were upregulated in intractable cancers. Genes that encode proteins implicated in the control of DNA transcription were downregulated in the intractable cancers. We describe hypotheses with regard to cell functions that may help in designing new therapeutic modalities, aimed at improving survival of cancer patients.

Two vaccine toxicity-related genes Agp and Hpx could prove useful for pertussis vaccine safety control

Conventional animal tests such as leukocytosis promoting tests have been used for decades to evaluate toxicity of pertussis vaccine. Here, we examined gene expression in relation to the vaccine toxicity using a DNA microarray. Comparison of conventional animal test data with the DNA microarray-based gene expression data revealed a gene expression pattern highly correlated with leukocytosis in animals. Of 10,490 rat genes analyzed, two genes, α1-acid-glycoprotein (Agp) and hemopexin (Hpx), were found up-regulated by the toxin administration in a dose-dependent manner (assayed by a quantitative PCR based on the microarray). Variation of the gene expression was very small amongst the test animals, and the results were highly reproducible. These findings suggest that gene expression analysis of vaccine-treated animals can be used as an accurate and simple method of pertussis vaccine safety assessment.

Development and application of an integrated drug‐drug interaction database focused on drug‐metabolizing P450 enzymes

Drug-drug interactions (DDIs) caused by direct chemical inhibition of key drug-metabolizing cytochrome P450 enzymes (CYPs) by a co-administered drug have been well documented and well understood. However, many other well-documented DDIs can not be so readily explained. Recent investigations into drug and other xenobiotic-mediated expression changes of CYP genes have broadened our understanding of drug metabolism and DDI. In order to gain additional information on DDI, we have integrated existing information on drugs that are substrates, inhibitors, or inducers of important drug-metabolizing CYPs with new data on drug-mediated expression changes of the same set of CYPs from DrugMatrix®, a large-scale microarray gene expression database of drug-treated rat tissues. The integrated information has been organized into a drug-cytochrome P450 matrix which facilitates comparative analysis of drugs that are P450 substrates against drugs that are P450 inhibitors or inducers/suppressors of CYP gene expression. When examined at the gene expression level, 119 currently marketed drugs from 265 studied were CYP inducers and 83 were suppressors. Information in this drug-drug interaction database allows the user to consider drug-mediated CYP-suppression as a mechanism for previously well documented but not well understood DDIs. The value of this new information will be illustrated by a number of examples.

AthaMap web tools for the analysis and identification of co-regulated genes

The AthaMap database generates a map of cis-regulatory elements for the whole Arabidopsis thaliana genome. This database has been extended by new tools to identify common cis-regulatory elements in specific regions of user-provided gene sets. A resulting table displays all cis-regulatory elements annotated in AthaMap including positional information relative to the respective gene. Further tables show overviews with the number of individual transcription factor binding sites (TFBS) present and TFBS common to the whole set of genes. Over represented cis-elements are easily identified. These features were used to detect specific enrichment of drought-responsive elements in cold-induced genes. For identification of co-regulated genes, the output table of the colocalization function was extended to show the closest genes and their relative distances to the colocalizing TFBS. Gene sets determined by this function can be used for a co-regulation analysis in microarray gene expression databases such as Genevestigator or PathoPlant. Additional improvements of AthaMap include display of the gene structure in the sequence window and a significant data increase. AthaMap is freely available at .

A fisheye viewer for microarray-based gene expression data

BackgroundMicroarray has been widely used to measure the relative amounts of every mRNA transcript from the genome in a single scan. Biologists have been accustomed to reading their experimental data directly from tables. However, microarray data are quite large and are stored in a series of files in a machine-readable format, so direct reading of the full data set is not feasible. The challenge is to design a user interface that allows biologists to usefully view large tables of raw microarray-based gene expression data. This paper presents one such interface – an electronic table (E-table) that uses fisheye distortion technology.ResultsThe Fisheye Viewer for microarray-based gene expression data has been successfully developed to view MIAME data stored in the MAGE-ML format. The viewer can be downloaded from the project web site . The fisheye viewer was implemented in Java so that it could run on multiple platforms. We implemented the E-table by adapting JTable, a default table implementation in the Java Swing user interface library. Fisheye views use variable magnification to balance magnification for easy viewing and compression for maximizing the amount of data on the screen.ConclusionThis Fisheye Viewer is a lightweight but useful tool for biologists to quickly overview the raw microarray-based gene expression data in an E-table.

Spotting and validation of a genome wide oligonucleotide chip with duplicate measurement of each gene

The quality of DNA microarray based gene expression data relies on the reproducibility of several steps in a microarray experiment. We have developed a spotted genome wide microarray chip with oligonucleotides printed in duplicate in order to minimise undesirable biases, thereby optimising detection of true differential expression. The validation study design consisted of an assessment of the microarray chip performance using the MessageAmp and FairPlay labelling kits. Intraclass correlation coefficient (ICC) was used to demonstrate that MessageAmp was significantly more reproducible than FairPlay. Further examinations with MessageAmp revealed the applicability of the system. The linear range of the chips was three orders of magnitude, the precision was high, as 95% of measurements deviated less than 1.24-fold from the expected value, and the coefficient of variation for relative expression was 13.6%. Relative quantitation was more reproducible than absolute quantitation and substantial reduction of variance was attained with duplicate spotting. An analysis of variance (ANOVA) demonstrated no significant day-to-day variation.

Handling missing DNA microarray data by kriging estimators

Microarray gene expression data provide life science researchers with much more sensitive and detailed information about gene expression patterns than conventional methodologies for the purpose of facilitating gene recognition efforts. However, due to insufficient image resolution and noise generated during microarray experiments, gene expression matrices are frequently represented with missing elements. Methods for estimating missing microarray data are therefore needed to allow further analysis. In this paper, we present two kriging estimators for estimating missing values in DNA microarrays. These approaches can be useful for downstream analysis of microarray-based gene expression data.

Drug-induced changes in P450 enzyme expression at the gene expression level: A new dimension to the analysis of drug–drug interactions

Drug–drug interactions (DDIs) caused by direct chemical inhibition of key drug-metabolizing cytochrome P450 enzymes by a co-administered drug have been well documented and well understood. However, many other well-documented DDIs cannot be so readily explained. Recent investigations into drug and other xenobiotic-mediated expression changes of P450 genes have broadened our understanding of drug metabolism and DDI. In order to gain additional information on DDI, we have integrated existing information on drugs that are substrates, inhibitors, or inducers of important drug-metabolizing P450s with new data on drug-mediated expression changes of the same set of cytochrome P450s from a large-scale microarray gene expression database of drug-treated rat tissues. Existing information on substrates and inhibitors has been updated and reorganized into drug–cytochrome P450 matrices in order to facilitate comparative analysis of new information on inducers and suppressors. When examined at the gene expression level, a total of 119 currently marketed drugs from 265 examined were found to be cytochrome P450 inducers, and 83 were found to be suppressors. The value of this new information is illustrated with a more detailed examination of the DDI between PPARα agonists and HMG-CoA reductase inhibitors. This paper proposes that the well-documented, but poorly understood, increase in incidence of rhabdomyolysis when a PPARα agonist is co-administered with a HMG-CoA reductase inhibitor is at least in part the result of PPARα-induced general suppression of drug metabolism enzymes in liver. The authors believe this type of information will provide insights to other poorly understood DDI questions and stimulate further laboratory and clinical investigations on xenobiotic-mediated induction and suppression of drug metabolism.