Large-Scale Gene Expression Data Analysis: A New Challenge to Computational Biologists

Gene expression profiling of mouse aborted uterus induced by lipopolysac charide

To identify genes that participate in the abortion process, normal pregnant uteri were compared to lipopolysaccharide (LPS)-induced abortion uteri. At day 6 of pregnancy, mice were treated with LPS at various time points to induce an abortion. Total RNAs were applied to a cDNA microarray to analyze genes with altered expression. At the early stage (2 hours) of LPS-induced abortion, upregulated genes were mainly composed of immune responsive genes, including Ccl4, Ccl2, Cxcl13, Gbp3, Gbp2, Mx2, H2-Eb1, Irf1 and Ifi203. Genes related to toll-like receptor signaling were also overexpressed. At late stages of abortion (12-24 hours), many genes were suppressed rather than activated, and these were mainly related to the extracellular matrix, cytoskeleton, and anti-apoptosis. Altered expression of several selected genes was confirmed by real time reverse transcription-polymerase chain reaction. The results demonstrated that many known genes were altered in the LPS-treated pregnant uterus, implying that the molecular mechanisms of the genes involved in LPS-induced abortion are complicated. Further analysis of this expression profile will help our understanding of the pathophysiological basis for abortion.

Molecular diagnosis and prognosis with DNA microarrays

Microarray analysis makes it possible to determine thousands of gene expression values simultaneously. Changes in gene expression, as a response to diseases, can be detected allowing a better understanding and differentiation of diseases at a molecular level. By comparing different kinds of tissue, for example healthy tissue and cancer tissue, the microarray analysis indicates induced gene activity, repressed gene activity or when there is no change in the gene activity level. Fundamental patterns in gene expression are extracted by several clustering and machine learning algorithms. Certain kinds of cancer can be divided into subtypes, with different clinical outcomes, by their specific gene expression patterns. This enables a better diagnosis and tailoring of individual patient treatments.

Public consortium efforts in toxicogenomics

Public consortia provide a forum for addressing questions requiring more resources than one organization alone could bring to bear and engaging many sectors of the scientific community. They are particular well suited for tackling some of the questions encountered in the field of toxicogenomics, where the number of studies and microarray analyses would be prohibitively expensive for a single organization to carry out. Five consortia that stand out in the field of toxicogenomics are the Institutional Life Sciences Institute (ILSI) Health and Environmental Sciences Institute (HESI) Committee on the Application of Genomics to Mechanism Based Risk Assessment, the Toxicogenomics Research Consortium, the MicroArray Quality Control (MAQC) Consortium, the InnoMed PredTox effort, and the Predictive Safety Testing Consortium. Collectively, these consortia efforts have addressed issues such as reproducibility of microarray results, standard practice for assays and analysis, relevance of microarray results to conventional end points, and robustness of statistical models on diverse data sets. Their results demonstrate the impact that the pooling of resources, experience, expertise, and insight found in consortia can have.

Expression profiling of muscles from Fukuyama-type congenital muscular dystrophy and laminin-α2 deficient congenital muscular dystrophy; is congenital muscular dystrophy a primary fibrotic disease?

Fukuyama-type congenital muscular dystrophy (FCMD) and laminin-alpha2 deficient congenital muscular dystrophy (MDC1A) are congenital muscular dystrophies (CMDs) and they both are categorized into the same clinical entity of muscular dystrophy as Duchenne muscular dystrophy (DMD). All three disorders share a common etiologic defect in the dystrophin-glycoprotein complex, which connects muscle structural proteins with the extracellular basement membrane. To investigate the pathophysiology of these CMDs, we generated microarray gene expression profiles of skeletal muscle from patients in various clinical stages. Despite diverse pathological changes, the correlation coefficient of overall gene expression among these samples was considerably high. We performed a multi-dimensional statistical analysis, the Distillation, to extract determinant genes that distinguish CMD muscle from normal controls. Up-regulated genes were primarily extracellular matrix (ECM) components, whereas down-regulated genes included structural components of mature muscle. These observations reflect active interstitial fibrosis with less active regeneration of muscle cell components in the CMDs, characteristics that are clearly distinct from those of DMD. Although the severity of fibrosis varied among the specimens tested, ECM gene expression was consistently high without substantial changes through the clinical course. Further, in situ hybridization showed more prominent ECM gene expression on muscle cells than on interstitial tissue cells, suggesting that ECM components are induced by regeneration process rather than by 'dystrophy.' These data imply that the etiology of FCMD and MDC1A differs from that of the chronic phase of classical muscular dystrophy, and the major pathophysiologic change in CMDs might instead result from primary active fibrosis.

St John's wort and imipramine-induced gene expression profiles identify cellular functions relevant to antidepressant action and novel pharmacogenetic candidates for the phenotype of antidepressant treatment response

Both the prototypic tricyclic antidepressant imipramine (IMI) and the herbal product St John's wort (SJW) can be effective in the treatment of major depressive disorder. We studied hypothalamic gene expression in rats treated with SJW or IMI to test the hypothesis that chronic antidepressant treatment by various classes of drugs results in shared patterns of gene expression that may underlie their therapeutic effects. Individual hypothalami were hybridized to individual Affymetrix chips; we studied three arrays per group treatment. We constructed 95% confidence intervals for expression fold change for genes present in at least one treatment condition and we considered genes to be differentially expressed if they had a confidence interval excluding 1 (or -1) and had absolute difference in expression value of 10 or greater. SJW treatment differentially regulated 66 genes and expression sequence tags (ESTs) and IMI treatment differentially regulated 74 genes and ESTs. We found six common transcripts in response to both treatments. The likelihood of this occurring by chance is 1.14 x 10(-23). These transcripts are relevant to two molecular machines, namely the ribosomes and microtubules, and one cellular organelle, the mitochondria. Both treatments also affected different genes that are part of the same cell function processes, such as glycolytic pathways and synaptic function. We identified single-nucleotide polymorphisms in the human orthologs of genes regulated both treatments, as those genes may be novel candidates for pharmacogenetic studies. Our data support the hypothesis that chronic antidepressant treatment by drugs of various classes may result in a common, final pathway of changes in gene expression in a discrete brain region.

Identification of Genetic Networks

In this report, we propose the use of structural equations as a tool for identifying and modeling genetic networks and genetic algorithms for searching the most likely genetic networks that best fit the data. After genetic networks are identified, it is fundamental to identify those networks influencing cell phenotypes. To accomplish this task we extend the concept of differential expression of the genes, widely used in gene expression data analysis, to genetic networks. We propose a definition for the differential expression of a genetic network and use the generalized T  2 statistic to measure the ability of genetic networks to distinguish different phenotypes. However, describing the differential expression of genetic networks is not enough for understanding biological systems because differences in the expression of genetic networks do not directly reflect regulatory strength between gene activities. Therefore, in this report we also introduce the concept of differentially regulated genetic networks, which has the potential to assess changes of gene regulation in response to perturbation in the environment and may provide new insights into the mechanism of diseases and biological processes. We propose five novel statistics to measure the differences in regulation of genetic networks. To illustrate the concepts and methods for reconstruction of genetic networks and identification of association of genetic networks with function, we applied the proposed models and algorithms to three data sets.

Identification of genes expressed with temporal-spatial restriction to developing cerebellar neuron precursors by a functional genomic approach

Hedgehog pathway activation is required for proliferation of cerebellar granule cell neuron precursors during development and is etiologic in certain cerebellar tumors. To identify genes expressed specifically in granule cell neuron precursors, we used oligonucleotide microarrays to analyze regulation of 13,179 genes/expressed sequence tags in heterogeneous primary cultures of neonatal mouse cerebellum that respond to the mitogen Sonic hedgehog. In conjunction, we applied experiment-specific noise models to render a gene-by-gene robust indication of up-regulation in Sonic hedgehog-treated cultures. Twelve genes so identified were tested, and 10 (83%) showed appropriate expression in the external granular layer (EGL) of the postnatal day (PN) 7 cerebellum and down-regulation by PN 15, as verified by in situ hybridization. Whole-organ profiling of the developing cerebellum was carried out from PN 1 to 30 to generate a database of temporal gene regulation profiles (TRPs). From the database an algorithm was developed to capture the TRP typical of EGL-specific genes. The "TRP-EGL" accurately predicted expression in vivo of an additional 18 genes/expressed sequence tags with a sensitivity of 80% and a specificity of 88%. We then compared the positive predictive value of our analytical procedure with other widely used methods, as verified by the TRP-EGL in silico. These findings suggest that replicate experiments and incorporation of noise models increase analytical specificity. They further show that genome-wide methods are an effective means to identify stage-specific gene expression in the developing granule cell lineage.

cis element/transcription factor analysis (cis/TF): a method for discovering transcription factor/cis element relationships

We report a simple new algorithm, cis/TF, that uses genomewide expression data and the full genomic sequence to match transcription factors to their binding sites. Most previous computational methods discovered binding sites by clustering genes having similar expression patterns and then identifying over-represented subsequences in the promoter regions of those genes. By contrast, cis/TF asserts that B is a likely binding site of a transcription factor T if the expression pattern of T is correlated to the composite expression patterns of all genes containing B, even when those genes are not mutually correlated. Thus, our method focuses on binding sites rather than genes. The algorithm has successfully identified experimentally-supported transcription factor binding relationships in tests on several data sets from Saccharomyces cerevisiae.

Towards environmental toxicogenomics -- development of a flow-through, high-density DNA hybridization array and its application to ecotoxicity assessment

Assessment of the environmental hazard posed by soils/sediments containing low to moderate levels of contaminants using standard analytical chemical methods is uncertain due (in part) to a lack of information on contaminant bioavailability, the unknown interactive effects of contaminant mixtures, our inability to determine the species of a metal in an environmental matrix, and the relative sensitivity of bioassay species. Regulatory agencies compensate for this uncertainty by lowering cleanup goals, but in this process they effectively exclude otherwise attractive cleanup options (i.e. bioremediation). Direct evaluations of soil and sediment toxicity preclude uncertainty from most of these sources. However, the time and cost of chronic toxicity tests limits their general application to higher levels of tiered toxicity assessments. Transcriptional level (mRNA) toxicity assessments offer great advantages in terms of speed, cost and sample throughput. These advantages are currently offset by questions about the environmental relevance of molecular level responses. To this end a flow-through, high-density DNA hybridization array (genosensor) system specifically designed for environmental risk assessment was developed. The genosensor is based on highly regular microchannel glass wafers to which gene probes are covalently bound at discrete (200-microm diameter spot) and addressable (250-microm spot pitch) locations. The flow-through design enables hybridization and washing times to be reduced from approximately 18 h to 20 min. The genosensor was configured so that DNA from 28 environmental samples can be simultaneously hybridized with up to 64 different gene probes. The standard microscopic slide format facilitates data capture with most automated array readers and, thus high sample throughput (> 350 sample/h). In conclusion, hardware development for molecular analysis is enabling very tractable means for analyzing RNA and DNA. These developments have underscored the need for further developmental work in probe design software, and the need to relate transcriptional level data to whole-organism toxicity indicators.

Computer-assisted identification of cell cycle-related genes: new targets for E2F transcription factors

The processes that take place during development and differentiation are directed through coordinated regulation of expression of a large number of genes. One such gene regulatory network provides cell cycle control in eukaryotic organisms. In this work, we have studied the structural features of the 5' regulatory regions of cell cycle-related genes. We developed a new method for identifying composite substructures (modules) in regulatory regions of genes consisting of a binding site for a key transcription factor and additional contextual motifs: potential targets for other transcription factors that may synergistically regulate gene transcription. Applying this method to cell cycle-related promoters, we created a program for context-specific identification of binding sites for transcription factors of the E2F family which are key regulators of the cell cycle. We found that E2F composite modules are found at a high frequency and in close proximity to the start of transcription in cell cycle-related promoters in comparison with other promoters. Using this information, we then searched for E2F sites in genomic sequences with the goal of identifying new genes which play important roles in controlling cell proliferation, differentiation and apoptosis. Using a chromatin immunoprecipitation assay, we then experimentally verified the binding of E2F in vivo to the promoters predicted by the computer-assisted methods. Our identification of new E2F target genes provides new insight into gene regulatory networks and provides a framework for continued analysis of the role of contextual promoter features in transcriptional regulation. The tools described are available at http://compel.bionet.nsc.ru/FunSite/SiteScan.html.

Automation for genomics, part two: sequencers, microarrays, and future trends

Automation for genomics has enabled a 43-fold increase in the total finished human genomic sequence in the world in the past four years. This is the second half of a two-part, noncomprehensive review that presents an overview of different types of automation equipment used in genome sequencing. The first part of the review, published in the previous issue, focused on automated procedures used to prepare DNA for sequencing or analysis. This second part of the review presents a look at available DNA sequencers and array technology and concludes with a look at future technologies. Alternate sequencing technologies including mass spectrometry, biochips, and single molecule analysis are included in this review.

Building a dictionary for genomes: identification of presumptive regulatory sites by statistical analysis

The availability of complete genome sequences and mRNA expression data for all genes creates new opportunities and challenges for identifying DNA sequence motifs that control gene expression. An algorithm, "MobyDick," is presented that decomposes a set of DNA sequences into the most probable dictionary of motifs or words. This method is applicable to any set of DNA sequences: for example, all upstream regions in a genome or all genes expressed under certain conditions. Identification of words is based on a probabilistic segmentation model in which the significance of longer words is deduced from the frequency of shorter ones of various lengths, eliminating the need for a separate set of reference data to define probabilities. We have built a dictionary with 1,200 words for the 6, 000 upstream regulatory regions in the yeast genome; the 500 most significant words (some with as few as 10 copies in all of the upstream regions) match 114 of 443 experimentally determined sites (a significance level of 18 standard deviations). When analyzing all of the genes up-regulated during sporulation as a group, we find many motifs in addition to the few previously identified by analyzing the subclusters individually to the expression subclusters. Applying MobyDick to the genes derepressed when the general repressor Tup1 is deleted, we find known as well as putative binding sites for its regulatory partners.

Genes for cognitive function: developments on the X

Developments in human genome research enabled the first steps toward a molecular understanding of cognitive function. That there are numerous genes on the X chromosome affecting intelligence at the lower end of the cognitive range is no longer in doubt. Naturally occurring mutations have so far led to the identification of seven genes accounting for a small proportion of familial nonspecific X-linked mental retardation. These new data indicate that normal expression of many more X-linked and autosomal genes contribute to cognitive function. The emerging knowledge implicating genes in intracellular signaling pathways provides the insight to identify as candidates other X-linked and autosomal genes regulating the normal development of cognitive function. Recent advances in unravelling the underlying molecular complexity have been spectacular but represent only the beginning, and new technologies will need to be introduced to complete the picture.