Rozprawy doktorskie na temat „Gene expression analysis”

This thesis describes the development of the application of modular regulation analysis, a subset of metabolic control analysis to microarray data. Microarray experiments measure complex changes in the abundance of many mRNAs under different conditions. Current analysis methods, such as clustering, cannot distinguish between direct and indirect effects on expression, or calculate the relative importance of mRNAs in effecting responses. Modular regulation analysis of microarray data reveals and quantifies which mRNA changes are important for cellular responses.  The mRNAs are clustered, then how perturbations alter each cluster (integrated response co-efficients) and how strongly those clusters affect an output response is calculated (elasticity co-efficients). The product of these values quantifies how an input changes a response through each cluster (partial response co-efficients). Once identified, important clusters that mediate a large proportion of the response may suggest targets for investigation of, for example, disease mechanisms, and way of modifying that response, such as potential knockout, overexpression or drug targets. Two published datasets were used throughout the development of the method. This determined the requirements of a suitable dataset, and involved the creation of a test to exclude problematic experiments from the dataset. Analyses of the two datasets using the final method reveal that two mRNA clusters transmit most of the response of yeast doubling time to galactose; one contains mainly galactose metabolic genes, and the other a regulatory gene. Analysis of the response of yeast relative fitness to 2-deocy-D-glucose reveals that control is distributed between several mRNA clusters. Monte Carlo analysis revealed that the co-efficients were not statistically significant, due to the large amount of experimental error in the dataset. However, modular regulation analysis should become more applicable in practice as microarray technology is improving rapidly.

Breast cancer is a leading cause of malignancy worldwide. Improvements to gene expression profiling technology have resulted in the identification of many prognostic and predictive gene expression signatures for breast cancer. Whilst some of these signatures are being developed commercially, only two prognostic signatures: MammaPrint and Oncotype DX, are currently being validated in clinical trials. Many of these gene expression signatures require independent validation and the underlying biology behind these signatures remains unclear. The aim of this thesis is to identify key molecular pathways that are relevant in breast cancer using pathway and network analyses of new and existing in vivo and in vitro microarray gene expression data. The oestrogen signalling pathway was the main focus of this thesis due to its documented importance in the pathogenesis of breast cancer. Analyses of gene expression differences in New Zealand breast tumours according to oestrogen receptor (ER) status revealed differentially regulated genes such as ESR1, GATA3 and EGFR, which have also been reported in other breast cancer microarray studies. The analyses of a collaboratively assembled 960-tumour dataset of clinical breast cancer microarray data revealed differentially regulated pathways involving BCL2, ESR1, EGFR, MYC and NFKB between ER positive and negative tumours. We also identified a principal component of oestrogen activity using the gene expression data from our collated 960-tumour dataset, that could be used alongside ESR1 mRNA and ER protein expression (from immunohistochemistry) to stratify breast cancer patients more accurately. The generation of an in vitro siRNA perturbation dataset using MCF7 breast cancer cells, and its analyses using gene networks has identified relationships between genes that appear to operate both in vitro and in vivo. There were more highly correlated gene pairs shared between the MCF7 dataset and luminal A tumours than between this dataset and other tumour subtypes. The identification of key molecular pathways and master regulators operating in breast tumours from gene expression data may improve our understanding of the biology behind breast cancer. This knowledge can be used in the future to help integrate gene expression data with clinicohistopathological data to improve diagnostic and therapeutic decision-making for patients.

Combining effect sizes from individual studies using random-effects models are commonly applied in high-dimensional gene expression data. However, unknown study heterogeneity can arise from inconsistency of sample qualities and experimental conditions. High heterogeneity of effect sizes can reduce statistical power of the models. We proposed two new methods for random effects estimation and measurements for model variation and strength of the study heterogeneity. We then developed a statistical technique to test for significance of random effects and identify heterogeneous genes. We also proposed another meta-analytic approach that incorporates informative weights in the random effects meta-analysis models. We compared the proposed methods with the standard and existing meta-analytic techniques in the classical and Bayesian frameworks. We demonstrate our results through a series of simulations and application in gene expression neurodegenerative diseases.

This thesis developed and applied Bayesian models for the analysis of survival data. The gene expression was considered as explanatory variables within the Bayesian survival model which can be considered the new contribution in the analysis of such data. The censoring factor that is inherent of survival data has also been addressed in terms of its impact on the fitting of a finite mixture of Weibull distribution with and without covariates. To investigate this, simulation study were carried out under several censoring percentages. Censoring percentage as high as 80% is acceptable here as the work involved high dimensional data. Lastly the Bayesian model averaging approach was developed to incorporate model uncertainty in the prediction of survival.

Telomere shortening and an increased telomerase activity are associated with poor prognosis and disease progression in many cancers. In Chronic Myeloid Leukaemia (CML) telomere shortening has a strong correlation with disease progression. Expression of hTERT, the catalytic component of telomerase, was evaluated in the CD34+ cells of CML patients. This revealed that expression of hTERT was significantly reduced in chronic phase CML and decreased with disease progression to accelerated phase and blast crisis. .It could therefore be concluded that reduced hTERT expression contributes to reduced telomere length in CML. Additionally, expression of c-Myc, which increases hTERT transcription, correlated with hTERT expression suggesting decreased hTERT is partly caused by reduced c-Myc. hTERT promoter methylation and mutation status were investigated and this revealed that the hTERT promoter was not methylated and mutation rates were low suggesting that these are not contributing to reduced hTERT expression. cDNA microarrays were used to analyse gene expression in neutrophils of patients with Essential Thrombocythaemia (ET) harbouring the JAK2 V617F mutation which, like the BCRlABL translocation in CML, results in an activated kinase. Neutrophils of ET patients exhibited a gene expression profile close to that of controls despite the presence of the mutation. Affymetrix microarrays were used to investigate the role of telomerase related genes in Myelodysplastic Syndromes (MDS). Genes decreased in patients with del(5q) include positive regulators of telomere length and genes with higher expression were associated with increased telomerase activity.

Thesis (M.Phil.)--City University of Hong Kong, 2005.
"Submitted to Department of Computer Engineering and Information Technology in partial fulfillment of the requirements for the degree of Master of Philosophy" Includes bibliographical references (leaves 70-79)

Gene expression is random owing to the low copy numbers of molecules in a living cell and the best way to study it is by use of a stochastic method, specifically the chemical master equation. The method is used here to derive analytically the invariant probability distributions, and expressions for the moments and noise strength for a simple gene model without feedback. Sensitivity analysis, emphasizing particularly the dependence of the probability distributions, the moments, and noise strength is carried out using Metabolic Control Analysis, which uses control coefficients that measure the response of observables when parameters change. Bifurcation analysis is also carried out. The results show that the number of mRNA molecules follows a hypergeometric probability distribution, and that noise decreases as the number of these molecules increases. Metabolic Control Analysis was successfully extended to genetic control mechanisms, with the obtained control coefficients satisfying a summation theorem. The system undergoes stochastic bifurcations as parameters change.
xii, 86 leaves : ill. ; 29 cm

Mutation of two NMDA receptor complex proteins, PSD95 and SynGAP, results in altered phenotypes in NMDA dependent phenomena. As components of the receptor complex, necessary for NMDA mediated signalling, mutation of these proteins may alter pathways that regulate gene expression. Affymetrix microarray analysis of RNA from PSD95^-/- forebrain and SynGAP^-/- hippocampi, compared to wildtype samples, revealed significant changes, greater than 1.5 fold, in a limited numbers of genes. Of the 12000 transcripts analysed 0.22% were significantly altered in PSD95 mutant tissue and 0.35% were changed in SynGAP mutant tissue. The genes altered in each genotype were distinct, apart from an overlap of 3 genes that were found similarly down regulated in PSD95^-/- forebrain and SynGAP^-/- hippocampi. These 3 genes, c-fos, nur77 and egr2, are activity dependent and are regulated, in part, through the NMDA receptor. It is possible that changes in gene expression may underlie the electrophysiological and behavioural phenotypes seen PSD95^-/- and SynGAP^+/- animals. It is likely that the genes altered in each of the mutants represent a subset of the genes regulated by NMDA receptor signalling. To get an understanding of the complete set of genes regulated by the NMDA receptor complex and in vitro method of NMDA receptor stimulation was sought. A primary cultured neuron system was used, allowing NMDA receptor activity to be manipulated by the pharmacological treatment of the culture medium. NMDA and bicuculline treatment of primary cultured cortical neurons proved ineffective methods of inducing activity dependent genes as measured by cfos expression. However, AP5 treatment of primary cultured neurons decreased activity dependent gene expression. Electrophysiological analysis of the cultures revealed that bicuculline treatment had no significant effect on culture activity, where as AP5 treatment caused a significant decrease in neuronal activity.

We use graph theory to model a database of gene expression levels and provide a tool that can assist in designing biological experiments which could help to better understand the interactions between genes. The data was provided by StemBase (http://www.stembase.ca). We model a portion of this database as a graph, and study some parameters on it that may be biologically relevant. We include theoretical discussion of the parameters, and presentation of algorithms for their computation. The focus is on structural properties of the graph; thus, we investigate the graph's bipartiteness, connected components, distance between vertices, radius, diameter, center, cut-vertices and blocks, and its maximal and maximum cliques. We develop the terminology and results used for modeling the database, and implement some of the algorithms in MATLAB. We analyze the results of running the code on the database and discuss their relevance.

Cellular senescence is an irreversible growth-arrested state seen in primary human cells after a finite number of cell divisions both in culture and in vivo. The escape from senescence to immortalisation is often thought to be a prerequisite for carcinogenesis. Many changes in senescent cells are consistent with changes in tissue and organ function with age. I used serial analysis of gene expression (SAGE) to analyse global gene expression profiles in senescent and early passage human foetal fibroblasts (hff). A total of over 20,000 SAGE tags were sequenced and characterised, corresponding to 2,675 unique transcripts. Relative to the early passage hff, transcripts which were found to be markedly increased in senescent hff included those encoding p21^WAF1, Cyclin D1, ferritin heavy chain, transforming growth factor-beta induced gene (BIGH3), skin collagenase and amyloid. Amongst them, genes such as skin collagenase and amyloid are known to be up-regulated in human ageing. Together with these known genes, a number of "unknown" genes were up-regulated. Seven differentially expressed genes, up-regulated in senescence (BIGH3, prion, dickkopf (Xenopus laevis) homolog 1 (dkk1), Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 2 (CITED2), Rap1a and Cysteine knot superfamily 1, bone morphogenetic protein antagonist 1 and retinoic acid receptor, alpha (RARA)) were selected for validation with reverse transcriptase-polymerase chain reaction (RT-PCR). The roles of these genes are discussed in relation to cancer and neurodegenerative diseases. My study demonstrates that replicative senescence is a complex phenomenon involving genes from the wingless-type (Wnt), mitogen-activated protein (Map) kinase kinases extracellular signal-regulated (Erk) kinase, transforming growth factor beta signalling and epigenetic pathways. The challenges remaining now are to identify senescence-related tumour suppressor gene(s) and to elucidate further the biochemical pathways of senescence.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Chemical Engineering, 2004.
Includes bibliographical references (p. 65-66).
Gene expression arrays are a technology used to measure quantities of messenger ribonucleic acid (mRNA). Application of the technology involves a variety of physical processes beginning with the acquisition of mRNA samples and ending with the fluorescence imaging of a gene expression array. This thesis examines these physical processes, develops a mechanistic model, and derives the analysis procedure based on the model. Chief advantages of this approach are that it accounts for certain previously unexplained array phenomena and is based in a clear way on physical knowledge allowing non-arbitrary determination of both the probability that any given gene has altered expression ratio relative to a control as well as the magnitude of this induction or repression. We demonstrate its use on simulated and real array data, and show that a considerable amount of previously unrecognized information concerning gene expression differences is inherent in the array measurements.
by Keith Howard Duggar.
Ph.D.

Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2004.
Includes bibliographical references (p. 68-71).
A method is proposed for gene expression analysis relying upon convex matrix factorization (CMF). In CMF, one of the matrix factors has a convexity constraint, that is, each row is nonnegative and sums to one, and hence can be interpreted as a probability distribution. This is motivated biologically by expression data resulting from a mixture of different cell types. This thesis investigates implementing CMF with various constraints applied to the expression matrix, and applies the technique to a problem in analysis of the cell cycle and two problems in cancer classification.
by John D. Barnett.
S.M.

Breast Cancer is the most commonly diagnosed cancer in women, with more than 1.2 million women diagnosed annually worldwide. It is also a frequently fatal disease and remains difficult to treat, despite advances in all facets of cancer management. While a number of genetic mutations have been identified in human breast cancers, he specific combinations of the mutations required in concert for formation of a breast carcinoma remains unknown, making precise detection or prognostic predictions impossible. Although estrogen receptor (ER) status is predictive of response to hormonal treatments, there are currently no clinically useful predictive markers of a patient’s response to chemotherapy. This results in all patients who are eligible for chemotherapy receiving the same treatment even though de novo drug resistance will result in the treatment failing in about 80% of cases. Developing improved diagnostic tools to cluster different breast cancers into groups based on genetic parameters has the potential to revolutionise individualised treatment options and subsequent efficacy. This in turn will improve quality of life for patients undergoing therapy who will no longer suffer the consequences of unnecessary treatments and more importantly, a subsequent improved survival rate.
Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Medical Science
Griffith Health
Full Text

Immobilization is an effective method to increase ethanol production, as proven by previous research. Results almost exclusively demonstrate an increase in ethanol production by and decrease in reproduction rate of immobilized Saccharomyces cerevisiae cells. Recently, research has been conducted to determine the cause of this change. The extreme variance in results due to lack of technology makes it difficult to determine the cellular changes induced by immobilization. With the advent of new technology, specifically gene expression analysis, the RNA content of cells can be easily and rapidly analyzed. S. cerevisiae cells were immobilized in 3% (w/v) calcium alginate beads and grown inside of a packed bed reactor for comparison to planktonic cells growing in batch and chemostat cultures. Temperature inside of the reactor was maintained at 33 C with a pH of 5.5. Cell concentration inside of the beads was monitored periodically in order to create growth curves. Bud scar numbers of immobilized cells were also counted and compared to suspended cells. Scanning electron microscopy images of the alginate beads were taken to determine cell growth inside of the beads. Affymetrix Yeast 2.0 gene chips were used, and the data retrieved was analyzed with GeneSpring software using the Bioconductor packages. Results indicated changes in expression of 3,559 genes with significant difference among treatments by a factor of 2-fold or greater. One-way ANOVA of the filtered data yielded 380 highly significantly different genes between immobilized and suspended cells. Many of the genes pertaining to glycolysis exhibited increased expression levels. Several genes necessary for reproduction were expressed at lower levels in the immobilized cells than in their planktonic counterparts. Many different gene ontologies are discussed, and the expressed genes are mapped onto biochemical pathways.

The use of comparative gene expression techniques has expanded considerably in recent years, especially with advances in microarray technology. In this project, a number of these techniques have been used to identify genes worthy of further research as potential mediators of bone cell differentiation and function. Bone is a tissue with many potential as yet unidentified regulatory molecules. The skeleton is constantly undergoing replacement, with old bone being degraded by osteoclasts, bone resorbing cells derived from the haematopoietic lineage, and replaced with new bone by osteoblasts, bone synthesizing cells derived from the mesenchymal stem cell lineage. When the rate of bone resorption exceeds the rate of bone synthesis, osteoporosis can occur. Osteoporosis is the most common form of disease affecting the skeleton, and one of the most common age-related diseases, and is a major social and economic burden. Recent studies have shown that cells of mesenchymal lineage are capable of adopting alternate differentiation fates, suggesting that cell-based therapies may be a useful therapeutic approach for this disease. Therefore, identification of molecular mechanisms involved in regulating the behavior and development of bone forming and bone resorbing cells is essential. The aim of this project, therefore, was to identify genes involved in various stages of bone cell differentiation using comparative gene expression techniques. The specific objectives of this project became: 1) to identify molecules expressed by osteoblasts which may increase or decrease bone synthesis; these may have potential for exploitation to treat bone loss in osteoporosis, or excess bone deposition in osteopetrosis; and 2) to identify molecules expressed by osteoclasts which may increase or decrease bone resorption; these may have potential for exploitation to treat excess bone deposition in osteopetrosis, or bone loss in osteoporosis. The first objective, identification of molecules expressed by osteoblasts involved in bone deposition, was addressed using three techniques: subtractive hybridization, DNA microarray analysis, and DNA macroarray analysis. These techniques were used to identify genes transcribed at different levels between foetal osteoblasts and fibroblasts. The key difference between DNA arrays, and subtractive hybridization, as techniques is that DNA arrays utilize a cDNA population fixed to a rigid medium as starting material. This means, therefore, that in order to identify a gene as being expressed, the gene must be present on the array, as a member of the cDNA library the original starting array was made from. This inhibits identification of truly novel transcripts, a bias which is removed in techniques such as subtractive hybridization. The technique of subtractive hybridization is used to identify genes transcribed at higher levels in one DNA sample compared with another. The subtractive hybridization technique described here was modified to enrich a foetal osteoblast phagemid library by removing phagemid which contain transcripts common to foetal osteoblasts and dermal fibroblasts, thus resulting in identification of genes expressed uniquely or at higher levels in osteoblasts. The technique identified 65 genes that were expressed either highly or specifically in osteoblasts when compared with fibroblasts. Some of the genes identified were found in multiple library clones, such as collagen and fibronectin, both of which are key structural components of bone, abundantly expressed by osteoblasts. Expression of some other identified genes had not previously been detected in osteoblasts, making them interesting targets for further investigation. Interesting genes revealed using this technique included prohibitin, leptin-receptor gene related protein, ornithine decarboxylase antizyme, amyloid precursor peptide and connective tissue growth factor. The usefulness of the technique was verified by performing real-time PCR to confirm the expression of these genes either specifically or abundantly in osteoblast cells. DNA array analysis was undertaken to identify transcripts previously identified in other tissues, but not investigated in bone cells to date. Micro and macroarray analysis was used to identify known genes that were over or underexpressed. The microarray comparison of fibroblasts and osteoblasts using cDNA differentially labeled with fluorescent dyes hybridized to glass microarrays, showed that the two cell types were very similar, with just 64 genes found to be regulated 5-fold or more- 37 were down-regulated in osteoblasts, and 27 were upregulated in osteoblasts, out of the 19,000 genes represented on the array. Genes shown to be significantly upregulated in osteoblasts by the 19k microarray included several neural proteins and transcription factors, while genes downregulated in osteoblasts included cell signal transducers and other transcriptional activity modifiers. The set of Atlas cDNA arrays consists of three pairs of arrays, with each pair containing 1176 genes. Two pairs of filters, or 2352 genes, were probed in the osteoblast/fibroblast comparison, while all three pairs, or 3528 genes, were probed in the osteoclast/macrophage comparison. The data from the DNA macroarray experiment, in which radioactively-labelled cDNA from osteoblasts and fibroblasts was hybridized to two separate sets of nylon arrays (ATLAS human cDNA arrays) showed that 12 out of the 2352 genes assayed were significantly regulated, with eight upregulated in osteoblasts, and four downregulated in osteoblasts. Genes upregulated in osteoblasts included transcription factor Dp-2, cAMP response element binding protein 1, and transcription factor ATF2. Genes down-regulated in osteoblasts included teratocarinoma derived growth factor, thymosin beta-10, and transcription factor 3. To address the second objective, and identify molecules expressed by osteoclasts involved in bone resorption, the DNA macroarray analysis approach was repeated. cDNA isolated from osteoclasts was compared with cDNA isolated from macrophages to identify genes differentially transcribed between these two cell types which differentiate from the same developmental lineage in vitro. DNA macroarray analysis, performed using radioactively labeled cDNA from osteoclasts and macrophages hybridized to ATLAS human cDNA arrays identified 53 genes as upregulated in osteoclasts, including GM-CSF Receptor, signalling molecule calmodulin 1, and transcription regulatory molecule Nuclear Factor of Activated T-cells (NFAT). Seventeen genes were shown to be downregulated in osteoclasts, including transcription factor zpf36, Integrin 5, and X-ray repair complementing protein. The studies described here suggested that osteoclasts and macrophages have more differentially expressed genes than osteoblasts and fibroblasts, with 1.98% of genes arrayed showing differential expression between osteoclasts and macrophages, but only 0.3% of genes arrayed showing differential expression between fibroblasts and osteoblasts. This suggests that the morphological differences between cell types may be a direct reflection of the molecular differences between them as well, as osteoblasts and fibroblasts are quite similar, while osteoclasts and macrophages are morphologically quite distinct. From this project, it would appear that comparisons of different populations of cells require the use of different techniques to yield the best results, and that the techniques of array analysis and subtractive hybridization may be best utilized in combination, rather than exclusively of each other. Many of the genes identified as differentially expressed between fibroblasts and osteoblasts using DNA arrays were fairly well-characterised 'house-keeping' type genes- metabolic, structural, and not specific or likely to play a significant role in the differentiation of osteoblasts, or the development of bone disease. One possible reason for this is that the arrays are too limited by the number of genes featured, to be able to detect many differences between similar cell types, where there are fewer differences to detect. In contrast, using the same macroarrays to compare the more distinct osteoclasts and macrophages resulted in identification of several interesting candidate genes, showing that some cell type comparisons can be performed adequately using this technology. By using the subtractive hybridization method to enrich a pre-existing phagemid library, any bias related to the genes able to be detected was removed. Although this technique requires manufacture of a cDNA library of the cell type of interest, it may be worthwhile for the comparison of similar cell types where sensitivity is an issue. In summary, this project used the techniques of subtractive hybridization and DNA macroarray and microarray analysis to detect genes showing differential expression between osteoblasts and fibroblasts, and used DNA macroarray analysis to detect genes differentially expressed between osteoclasts and macrophages. Of particular note were results for two interesting genes, amyloid precursor protein and ornithine decarboxylase antizyme. Amyloid Precursor Protein was identified as expressed in high levels in osteoblasts by subtractive hybridization, and real-time PCR studies later confirmed that it is expressed specifically in osteoblasts, and not at all in fibroblasts. Differential expression of ornithine decarboxylase antizyme was identified between both osteoclasts and macrophages, and fibroblasts and osteoblasts. Expression of antizyme results in destruction of ornithine decarboxylase, which is required for the production of polyamines. Degrading cells release spermidine, a polyamine, which attracts macrophages. It is possible that differential regulation of the inhibitory antizyme may be an important distinction between the function of macrophages as general tissue and debris scavenging cells, and osteoclasts, which specifically degrade bone. This study has identified some genes which further studies may show to be important regulators of cellular differentiation and behaviour.
Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Health Sciences
Full Text

The DNA microarray biotechnology simultaneously monitors the expression of thousands of genes and aims to identify genes that are differently expressed under different conditions. From the statistical point of view, it can be restated as identify genes strongly associated with the response or covariant of interest. The Gene Set Enrichment Analysis (GSEA) method is one method which focuses the analysis at the functional related gene sets level instead of single genes. It helps biologists to interpret the DNA microarray data by their previous biological knowledge of the genes in a gene set. GSEA has been shown to efficiently identify gene sets containing known disease-related genes in the real experiments. Here we want to evaluate the statistical power of this method by simulation studies. The results show that the the power of GSEA is good enough to identify the gene sets highly associated with the response or covariant of interest.

Made available in DSpace on 2014-06-12T15:59:06Z (GMT). No. of bitstreams: 2 arquivo4839_1.pdf: 1378221 bytes, checksum: f1a933734804959bb52fd2eef936641b (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2003
Large scale approaches, namely proteomics and transcriptomics, will play the most important role of the so-called post-genomics. These approaches allow experiments to measure the expression of thousands of genes from a cell in distinct time points. The analysis of this data can allow the the understanding of gene function and gene regulatory networks (Eisen et al., 1998). There has been a great deal of work on the computational analysis of gene expression time series, in which distinct data sets of gene expression, clustering techniques and proximity indices are used. However, the focus of most of these works are on biological results. Cluster validation has been applied in few works, but emphasis was given on the evaluation of the proposed validation methodologies (Azuaje, 2002; Lubovac et al., 2001; Yeung et al., 2001; Zhu & Zhang, 2000). As a result, there are few guidelines obtained by validity studies on which clustering methods or proximity indices are more suitable for the analysis of data from gene expression time series. Thus, this work performs a data driven comparative study of clustering methods and proximity indices used in the analysis of gene expression time series (or time courses). Five clustering methods encountered in the literature of gene expression analysis are compared: agglomerative hierarchical clustering, CLICK, dynamical clustering, k-means and self-organizing maps. In terms of proximity indices, versions of three indices are analysed: Euclidean distance, angular separation and Pearson correlation. In order to evaluate the methods, a k-fold cross-validation procedure adapted to unsupervised methods is applied. The accuracy of the results is assessed by the comparison of the partitions obtained in these experiments with gene annotation, such as protein function and series classification

Microarray technology allows measurement of the expression levels of thousand of genes simultaneously. Several gene set analysis (GSA) methods are widely used for extracting useful information from microarrays, for example identifying differentially expressed pathways associated with a particular biological process or disease phenotype. Though GSA methods like Gene Set Enrichment Analysis (GSEA) are widely used for pathway analysis, these methods are solely based on statistics. Such methods can be awkward to use if knowledge of specific pathways involved in particular biological processes are the aim of the study. Here we present a novel method (Knowledge Based Gene Set Analysis: KB-GSA) which integrates knowledge about user-selected pathways that are known to be involved in specific biological processes. The method generates an easy to understand graphical visualization of the changes in expression of the genes, complemented with some common statistics about the pathway of particular interest.

Today’s rapid development of powerful tools for geneexpression analysis provides unprecedented resources forelucidating complex molecular events.

The objective of this workhas been to apply, combine andevaluate tools for analysis of differential gene expressionusing atherosclerosis as a model system. First, an optimisedsolid-phase protocol for representational difference analysis(RDA) was applied to twoin vitromodel systems. Initially, The RDA enrichmentprocedure was investigated by shotgun cloning and sequencing ofsuccessive difference products. In the subsequent steps,combinations of RDA and microarray analysis were used tocombine the selectivity and sensitivity of RDA with thehigh-throughput nature of microarrays. This was achieved byimmobilization of RDA clones onto microarrays dedicated forgene expression analysis in atherosclerosis as well ashybridisation of labelled RDA products onto global microarrayscontaining more than 32,000 human clones. Finally, RDA wasapplied for the investigation of the focal localisation ofatherosclerotic plaques in mice usingin vivotissue samples as starting material.

A large number of differentially expressed clones wereisolated and confirmed by real time PCR. A very diverse rangeof gene fragments was identified in the RDA products especiallywhen they were screened with global microarrays. However, themicroarray data also seem to contain some noise which is ageneral problem using microarrays and should be compensated forby careful verification of the results.

Quite a large number of candidate genes related to theatherosclerotic process were found by these studies. Inparticular several nuclear receptors with altered expression inresponse to oxidized LDL were identified and deserve furtherinvestigation. Extended functional annotation does not liewithin the scope of this thesis but raw data in the form ofnovel sequences and accession numbers of known sequences havebeen made publicly available in GenBank. Parts of the data arealso available for interactive exploration on-line through aninteractive software tool. The data generated thus constitute abase for new hypotheses to be tested in the field ofatherosclerosis.

Keywords:representational difference analysis, geneexpression profiling, microarray analysis, atherosclerosis,foam cell formation

High-throughput gene expression data have been generated on a large scale by biologists.

The thesis describe a set of tools for the analysis of such data. It is specially gearded towards microarray data.
Doctorat en sciences appliquées
info:eu-repo/semantics/nonPublished

Various types of recurrent neural networks have been used as models for the regulatory relationships between genes. The neural network is trained on the data from micro-array techniques, each gene corresponds to a neuron in the network. The data from the micro-array technologies has numerous genes, but usually involves few samples, this makes the network heavily under-determined. In this work we will propose a method that can cope with the poorness of the data. We will use a Hopfield-type neural network to model the ontogenetic differentiation of female honeybees. A method that identifies the genes that determine the castes is proposed.

The aim of this research was to assemble sufficient experimental evidence about candidate gene transcript expression changes between non-neoplastic and neo- plastic colorectal tissues to justify future assay development involving promis- ing leads. To achieve this aim, this thesis explores the hypothesis that gene expression-based biomarkers can be used to accurately discriminate colorectal neoplastic tissues from non-neoplastic controls. This hypothesis was tested by first analysing multiple, large, quality controlled data sets comprising gene expression measurements across colorectal phenotypes to discover potential biomarkers. Candidate biomarkers were then subjected to validation testing using a custom-design oligonucleotide microarray applied to independently derived clinical specimens. A number of novel conclusions are reached based on these data. The most important conclusion is that a defined subset of genes expressed in the colorectal mucosa are reliably differentially ex- pressed in neoplastic tissues. In particular, the apparently high prediction accu- racy achieved for single gene transcripts to discriminate hundreds of neoplastic and non-neoplastic tissues provides compelling evidence that the resulting can- didate genes are worthy of further biomarker research. In addition to addressing the central hypothesis, additional contributions are made to the field of colorectal neoplasia gene expression profiling. These contributions include: The first systematic analysis of gene expression in non-diseased tissues along the colorectum To better understand the range of gene expression in non-diseased tissues, RNA extracts taken from along the longitudinal axis of the large intestine were studied. The development of quality control methodologies for high dimen- sional gene expression data Complex data collection platforms such as oligonucleotide microarrays introduce the potential for unrecognized confound- ing variables. The exploration of quality control parameters across five hundred microarray experiments provided insights about quality control techniques. The design of a custom microrray comprised of oligonucleotide probe- sets hybridising to RNA transcripts differentially expressed in neo- plastic colorectal specimens A custom design oligonucleotide microarray was designed and tested combining the results of multiple biomarker discovery projects. Introduction of a method to filter differentially expressed genes dur- ing discovery that may improve validation efficiencies of biomarker discovery based on gene expression measurements Differential expression discovery research is typically focused only on quantitative changes in transcript concentration between phenotype contrasts. This work introduces a method for generating hypotheses related to transcripts which may be quali- tatively “switched-on” between phenotypes. Identification of mRNA transcripts which are differentially expressed between colorectal adenomas and colorectal cancer tissues Transcripts differentially expressed between adenomatous and cancerous RNA extracts were discovered and then tested in independent tissues. In conclusion, these results confirm the hypothesis that gene expression profiling can discriminate colorectal neoplasia (including adenomas) from non-neoplastic controls. These results also establish a foundation for an ongoing biomarker development program.

Gene expression analysis has evolved as an extensive toolfor elucidation of various biological and molecular eventsoccurring in different organisms. A variety of techniques andsoftware tools have been developed to enable easier and morerapid means of exploring the genetic information. A moreeffective approach than exploring the whole content of genesexpressed under certain conditions is to study fingerprintassays or to use subtracted cDNA libraries to identify onlydifferentially expressed genes.

The objective for the work in this thesis has been toexplore differentially expressed genes in atherosclerosis. Thiswas done by applying and modifying a protocol for thesubtractive approach RDA (Representational Difference Analysis)in different model systems.

Initially, the molecular effects of an anti-atheroscleroticdrug candidate were elucidated. In addition, two alternativeapproaches to identify differentially expressed genes obtainedafter iterative rounds of RDA subtraction cycles wereevaluated. This revealed that in most cases, the shotgunapproach in which the obtained gene fragments are clonedwithout any prior selection has clear advantages compared tothe more commonly used selection strategy, whereby distinctbands are excised after gel electrophoresis.

A key process in the atherosclerotic plaque initiation isthe phenotypic change of macrophages into foam cells, which canbe triggered in a model system by using macrophages exposed tooxidised LDL. To investigate the genes expressed in thisprocess, the RDA technique was combined with microarrayanalysis, which allows for selectivity and sensitivity throughRDA, as well as rapid high-throughput analysis usingmicroarrays. The combination of these techniques enablessignificant differences in gene expression to be detected, evenfor weakly expressed genes and the results to be reliablyvalidated in a high throughput manner.

Finally, investigation of the focal nature ofatherosclerotic lesions and gene expression profiling werestudied using in vivo aortic tissues from ApoE-/- and LDLR -/-mice. The study was based on a comparison between localisationsthat are likely, and others that are unlikely, to developatherosclerotic plaques, and the RDA technique was employed toexplore differential gene expression.

Keywords:Representational Difference Analysis,atherosclerosis, gene expression profiling

Cardiovascular disease (CVD) causes 45% of on-duty firefighter fatalities, a high fraction even when compared to the risk of CVD found in other first-responder professions like police work and emergency medical services. Monitoring and managing firefighter cardiac health is important for both individual health and public safety. In our study, we are interested in assessing the utility of the most commonly used risk assessment scoring, known as the Framingham Risk Score, in evaluating the atherosclerotic risk in asymptomatic firefighters. To this end we determined FRS for 159 male firefighters from Gwinnett County, Georgia, and compared their risk categorization against their known atherosclerotic burden as determined by CIMT and CAC. While the 20% FRS threshold, corresponding to medium risk, had a high specificity for both CAC and CIMT, it also had a low sensitivity (17% and 40%, respectively), indicating that a large percentage of individuals with clinically significant atherosclerosis are being misclassified. By adjusting the FRS threshold downward, we were able to raise the sensitivity greatly with only a modest loss of specificity. Following percutaneous transluminal coronary angioplasty for the treatment of coronary artery disease, stents are commonly implanted at the treatment site to prevent recoil and negative remodeling. To combat in-stent restenosis, an arterial healing response that results in luminal loss in stented arteries, anti-restenotic drugs like sirolimus (SES) and zotarolimus (ZES) are commonly eluted by stents to suppress cell proliferation at the treatment site. While comparative studies have revealed significant difference between bare metal stents (BMS), SES, and ZES in both clinical and histological arterial response, the molecular basis of these differences remains poorly understood. We conducted a comparative gene expression profiling study using microarrays to examine differences in gene expression and pathway function in coronary arteries exposed to ZES, SES, and BMS in a porcine animal model. These molecular profiles suggest a model of delayed restenosis, resulting from a drug-induced suppression of inflammatory responses and proliferative processes, rather than an elimination of restenosis. microRNAs play a regulatory role in metastasis-related epithelial to mesenchymal transitions and mesenchymal to epithelial transitions in ovarian cancer cells. We previously showed that over-expression of miR-429 in ovarian cancer cells drove a transition from mesenchymal phenotypes to epithelial phenotypes both in morphology and expression of markers like ZEB1, ZEB2, and E-cadherin. Our study represents the first time course analysis of miR-429-induced MET in ovarian cancer cells. We transfected Hey cells with miR-429 and assayed gene expression over the course of 144 hours at regular intervals. The cell morphology and gene expression of our transfected cells changed to become more epithelial-like at 24 and 48 hours and then became more mesenchymal-like by 144 hours. By 144 hours the average gene expression levels for 98.6% of our genes were not significantly different from the levels they started from at 0 hours when we adjusted for baseline expression changes observed in our negative control treated cells. This suggests the use of microRNAs as cancer therapies and driving cancer cells to a more drug susceptible state.

BACKGROUND:Microarrays offer a powerful tool for diverse applications plant biology and crop improvement. Recently, two comprehensive assemblies of cotton ESTs were constructed based on three Gossypium species. Using these assemblies as templates, we describe the design and creation and of a publicly available oligonucleotide array for cotton, useful for all four of the cultivated species.RESULTS:Synthetic oligonucleotide probes were generated from exemplar sequences of a global assembly of 211,397 cotton ESTs derived from >50 different cDNA libraries representing many different tissue types and tissue treatments. A total of 22,787 oligonucleotide probes are included on the arrays, optimized to target the diversity of the transcriptome and previously studied cotton genes, transcription factors, and genes with homology to Arabidopsis. A small portion of the oligonucleotides target unidentified protein coding sequences, thereby providing an element of gene discovery. Because many oligonucleotides were based on ESTs from fiber-specific cDNA libraries, the microarray has direct application for analysis of the fiber transcriptome. To illustrate the utility of the microarray, we hybridized labeled bud and leaf cDNAs from G. hirsutum and demonstrate technical consistency of results.CONCLUSION:The cotton oligonucleotide microarray provides a reproducible platform for transcription profiling in cotton, and is made publicly available through http://cottonevolution.info webcite.

Bacteriophage Felix O1, also known as enterobacteria phage O1, has been used to type Salmonella Typhi and is an excellent candidate for use in bioremedial and therapeutic applications. It has extremely high intra-species specificity and is strictly virulent in nature, unable to undergo lysogeny. To facilitate the development of the bacteriophage for use in these areas, the full sequence of the genome had been elucidated previously. In this work, identification and classification of functional coding sequences via reverse transcriptase-polymerase chain reaction was performed.

All of the 115 putative open reading frames (ORFs) studied were found to be functional. 53.0%, 9.6%, and 18.3% of the ORFs investigated were found to initiate expression early, middle or late in the lytic cycle, respectively. Expression of the remaining 19.1% ORFs was evident when the amount of total RNA was increased and when samples were taken at a later time point. Comparisons between bacteriophage Felix O1 and the phage with the most shared homologs, phage T4, revealed many similarities in times of gene expression.
Master of Science

With the advent of microarray technology, it is possible to monitor gene expression of tens of thousands of genes in parallel. In order to gain useful biological knowledge, it is necessary to study the data and identify the underlying patterns, which challenges the conventional mathematical models. Clustering has been extensively used for gene expression data analysis to detect groups of related genes. The assumption in clustering gene expression data is that co-expression indicates co-regulation, thus clustering should identify genes that share similar functions. Microarray data contains plenty of uncertain and imprecise information. Fuzzy c-means (FCM) is an efficient model to deal with this type of data. However, it treats samples equally and cannot differentiate noise and meaningful data. In this thesis, motivated by the preservation of local structure, a local weighted FCM is proposed which concentrate on the samples in neighborhood. Experiments show that the proposed method is not only robust to the noise, but also identifies clusters with biological significance. Due to FCM is sensitive to the initialization and the choice of parameters, clustering result lacks stability and biological interpretability. In this thesis, a new clustering approach is proposed, which computes genes similarity in kernel space. It not only finds nonlinear relationship between gene expression profiles, but also identifies arbitrary shape of clusters. In addition, an initialization scheme is presented based on Parzen density estimation. The objective function is modified by adding a new weighted parameter, which accentuates the samples in high density areas. Furthermore, a parameters selection algorithm is incorporated with the proposed approach which can automatically find the optimal values for the parameters in the clustering process. Experiments on synthetic data and real gene expression data show that the proposed method substantially outperforms conventional models in term of stability and biological significance. Time series gene expression is a special kind of microarray data. FCM rarely consider the characteristics of the time series. In this work, a fuzzy clustering approach (FCMS) is proposed by using splines to smooth time-series expression profiles to minimize the noise and random variation, by which the general trend of expression can be identified. In addition, FCMS introduces a new geometry term of radius of curvature to capture the trend information between splines. Results demonstrate that the new method has substantial advantages over FCM for time-series expression data.

Kyoto University (京都大学)
0048
新制・課程博士
博士(農学)
甲第9781号
農博第1293号
新制||農||852(附属図書館)
学位論文||H14||N3712(農学部図書室)
UT51-2002-M159
京都大学大学院農学研究科応用生命科学専攻
(主査)教授 佐藤 文彦, 教授 關谷 次郎, 教授 泉井 桂
学位規則第4条第1項該当