Key aspects of analyzing microarray gene-expression data

One major challenge with the use of microarray technology is the analysis of massive amounts of gene-expression data for various applications. This review addresses the key aspects of the microarray gene-expression data analysis for the two most common objectives: class comparison and class prediction. Class comparison mainly aims to select which genes are differentially expressed across experimental conditions. Gene selection is separated into two steps: gene ranking and assigning a significance level. Class prediction uses expression profiling analysis to develop a prediction model for patient selection, diagnostic prediction or prognostic classification. Development of a prediction model involves two components: model building and performance assessment. It also describes two additional data analysis methods: gene-class testing and multiple ordering criteria.

Innovative proteomic approaches for cancer biomarker discovery

Substantial technological advances in proteomics and related computational science have been made in the past few years. These advances overcome in part the complexity and heterogeneity of the human proteome, permitting quantitative analysis and identification of protein changes associated with tumor development. Here, we discuss some of these advances that are uncovering new cancer biomarkers that have potential to detect cancer at early and curable stages and address remaining challenges.

Modeling cancer: integration of "omics" information in dynamic systems

The last 10 years have seen the rise of many technologies that produce an unprecedented amount of genome-scale data from many organisms. Although the research community has been successful in exploring these data, many challenges still persist. One of them is the effective integration of such data sets directly into approaches based on mathematical modeling of biological systems. Applications in cancer are a good example. The bridge between information and modeling in cancer can be achieved by two major types of complementary strategies. First, there is a bottom-up approach, in which data generates information about structure and relationship between components of a given system. In addition, there is a top-down approach, where cybernetic and systems-theoretical knowledge are used to create models that describe mechanisms and dynamics of the system. These approaches can also be linked to yield multi-scale models combining detailed mechanism and wide biological scope. Here we give an overall picture of this field and discuss possible strategies to approach the major challenges ahead.

MGED standards: work in progress

The Microarray Gene Expression Data (MGED) society is an international organization established in 1999 for facilitating sharing of functional genomics and proteomics array data. To facilitate microarray data sharing, the MGED society has been working in establishing the relevant data standards. The three main components (which will be described in more detail later) of MGED standards are Minimum Information About a Microarray Experiment (MIAME), a document that outlines the minimum information that should be reported about a microarray experiment to enable its unambiguous interpretation and reproduction; MAGE, which consists of three parts, The Microarray Gene Expression Object Model (MAGE-OM), an XML-based document exchange format (MAGE-ML), which is derived directly from the object model, and the supporting tool kit MAGEstk; and MO, or MGED Ontology, which defines sets of common terms and annotation rules for microarray experiments, enabling unambiguous annotation and efficient queries, data analysis and data exchange without loss of meaning. We discuss here how these standards have been established, how they have evolved, and how they are used.

Use of a 44k oligoarray to explore the transcriptome of Schistosoma mansoni adult worms

Recent advances in the study of Schistosoma mansoni genome and transcriptome have led to a better description of the S. mansoni gene complement. In this work, we report the design and use of a new S. mansoni 60-mer oligonucleotide microarray platform with approximately 44,000 probes, based on all publicly available cDNA sequence data for S. mansoni and Schistosoma japonicum. The large number of probes combined with the extensive sequence annotation available allowed a comprehensive approach, where most of the S. mansoni transcriptome is represented. Hybridization with adult worm RNA pointed to a set of genes transcriptionally active in this stage of the parasite's life cycle. Interestingly, a large proportion (43%) of genes for which transcription was detected in adults is comprised of "no match" genes, i.e. S. mansoni genes with unknown function and no identifiable orthologs in GenBank. Moreover, detection of bi-directional transcription for 7% of the active "no match" genes in adults leads us to hypothesize a widespread production of antisense RNA in S. mansoni, with possible regulatory roles.

Multiplexed microsphere diagnostic tools in gene expression applications: factors and futures

Microarrays have received significant attention in recent years as scientists have firstly identified factors that can produce reduced confidence in gene expression data obtained on these platforms, and secondly sought to establish laboratory practices and a set of standards by which data are reported with integrity. Microsphere-based assays represent a new generation of diagnostics in this field capable of providing substantial quantitative and qualitative information from gene expression profiling. However, for gene expression profiling, this type of platform is still in the demonstration phase, with issues arising from comparative studies in the literature not yet identified. It is desirable to identify potential parameters that are established as important in controlling the information derived from microsphere-based hybridizations to quantify gene expression. As these evolve, a standard set of parameters will be established that are required to be provided when data are submitted for publication. Here we initiate this process by identifying a number of parameters we have found to be important in microsphere-based assays designed for the quantification of low abundant genes which are variable between studies.

[Proteome analysis--basis for individualized pancreatic carcinoma therapy?]

Ductal pancreatic adenocarcinoma is a dismal disease, having the worst prognosis of all solid tumors. While genomics and transcriptomics have provided a wealth of data, no contribution has been made to clinical medicine in terms of diagnostic or prognostic markers. Hope lies in yet another novel technology, proteomics. Conceptually, proteomics bears the advantage of incorporating both posttranslational modifications as well as host factors. This is thought to be important in factors influencing survival such as chemoresistance. This tutorial review discusses the state of the art in pancreatic cancer proteomics in light of technical developments. At this moment, proteomics is still at the beginning in clinical application. First results, however, suggest some hope for the development of a new understanding of the molecular biology in pancreatic cancer yielding into very specific markers of disease or allowing a rational and individualized therapy.

Toxicogenomics of A375 human malignant melanoma cells

Toxicogenomics applications are increasingly applied to the evaluation of preclinical drug safety, and to explain toxicities associated with compounds at the mechanism level. In this review, we aim to describe the application of toxicogenomics tools for studying the genotoxic effect of active compounds on the gene-expression profile of A375 human malignant melanoma cells, through the other molecular functions of target genes, regulatory pathways and mechanisms of malignant melanomas. It also includes the current systems biology approaches, which are very useful for analyzing the biological system and understanding the entire mechanisms of malignant melanomas. We believe that this review would be very potent and useful for studying the toxicogenomics of A375 melanoma cells, and for further diagnostic and therapeutic applications.

Biological mechanisms of bone and cartilage remodelling--genomic perspective

Rapid advancements in the field of genomics, enabled by the achievements of the Human Genome Project and the complete decoding of the human genome, have opened an unimaginable set of opportunities for scientists to further unveil delicate mechanisms underlying the functional homeostasis of biological systems. The trend of applying whole-genome analysis techniques has also contributed to a better understanding of physiological and pathological processes involved in homeostasis of bone and cartilage tissues. Gene expression profiling studies have yielded novel insights into the complex interplay of osteoblast and osteoclast regulation, as well as paracrine and endocrine control of bone and cartilage remodelling. Mechanisms of new bone formation responsible for fracture healing and distraction osteogenesis, as well as healing of joint cartilage defects, have also been extensively studied. Microarray experiments have been especially useful in studying pathological processes involved in diseases such as osteoporosis or bone tumours. Existing results show that microarrays hold great promise in areas such as identification of targets for novel therapies or development of new biomarkers and classifiers in skeletal diseases.

Mining gene expression profiles: expression signatures as cancer phenotypes

Many examples highlight the power of gene expression profiles, or signatures, to inform an understanding of biological phenotypes. This is perhaps best seen in the context of cancer, where expression signatures have tremendous power to identify new subtypes and to predict clinical outcomes. Although the ability to interpret the meaning of the individual genes in these signatures remains a challenge, this does not diminish the power of the signature to characterize biological states. The use of these signatures as surrogate phenotypes has been particularly important, linking diverse experimental systems that dissect the complexity of biological systems with the in vivo setting in a way that was not previously feasible.

Current trends in computational inference from mass spectrometry-based proteomics

Mass spectrometry offers a high-throughput approach to quantifying the proteome associated with a biological sample and hence has become the primary approach of proteomic analyses. Computation is tightly coupled to this advanced technological platform as a required component of not only peptide and protein identification, but quantification and functional inference, such as protein modifications and interactions. Proteomics faces several key computational challenges such as identification of proteins and peptides from tandem mass spectra as well as their quantitation. In addition, the application of proteomics to systems biology requires understanding the functional proteome, including how the dynamics of the cell change in response to protein modifications and complex interactions between biomolecules. This review presents an overview of recently developed methods and their impact on these core computational challenges currently facing proteomics.

Current progress in network research: toward reference networks for key model organisms

The collection of multiple genome-scale datasets is now routine, and the frontier of research in systems biology has shifted accordingly. Rather than clustering a single dataset to produce a static map of functional modules, the focus today is on data integration, network alignment, interactive visualization and ontological markup. Because of the intrinsic noisiness of high-throughput measurements, statistical methods have been central to this effort. In this review, we briefly survey available datasets in functional genomics, review methods for data integration and network alignment, and describe recent work on using network models to guide experimental validation. We explain how the integration and validation steps spring from a Bayesian description of network uncertainty, and conclude by describing an important near-term milestone for systems biology: the construction of a set of rich reference networks for key model organisms.

The Schistosoma mansoni transcriptome: an update

Large scale EST sequencing projects have been carried out for Schistosoma mansoni and Schistosoma japonicum. This update will briefly review the most recent accomplishments in the area and discuss the use of EST data for the purposes of gene discovery, gene model development, genome annotation and SNP analysis. In addition, the use of ESTs for studying other features of the transcriptome such as splice site and transcription initiation variants will be discussed as well as approaches to assigning function to unknown transcripts. Although EST sequencing has contributed much for schistosome research, other data mining possibilities exist, including the identification of putative drug and vaccine targets.

Bayesian Inference of MicroRNA Targets from Sequence and Expression Data

MicroRNAs (miRNAs) regulate a large proportion of mammalian genes by hybridizing to targeted messenger RNAs (mRNAs) and down-regulating their translation into protein. Although much work has been done in the genome-wide computational prediction of miRNA genes and their target mRNAs, an open question is how to efficiently obtain functional miRNA targets from a large number of candidate miRNA targets predicted by existing computational algorithms. In this paper, we propose a novel Bayesian model and learning algorithm, GenMiR++ (Generative model for miRNA regulation), that accounts for patterns of gene expression using miRNA expression data and a set of candidate miRNA targets. A set of high-confidence functional miRNA targets are then obtained from the data using a Bayesian learning algorithm. Our model scores 467 high-confidence miRNA targets out of 1,770 targets obtained from TargetScanS in mouse at a false detection rate of 2.5%: several confirmed miRNA targets appear in our high-confidence set, such as the interactions between miR-92 and the signal transduction gene MAP2K4, as well as the relationship between miR-16 and BCL2, an anti-apoptotic gene which has been implicated in chronic lymphocytic leukemia. We present results on the robustness of our model showing that our learning algorithm is not sensitive to various perturbations of the data. Our high-confidence targets represent a significant increase in the number of miRNA targets and represent a starting point for a global understanding of gene regulation.

Compensating for Unknown Confounders in Microarray Data Analysis Using Filtered Permutations

Permutation of class labels is a common approach in microarray analysis. It is assumed to produce random score distributions, which are not affected by biological differences between samples. However, hidden confounding variables like the genetic background of patients or undetected experimental artifacts leave traces in the expression data contaminating the score distributions obtained from random permutations. While the effects of known confounders can be compensated using established methodology, little is known on how to deal with unknown confounders. We discuss a computational method called permutation filtering, which aims to borrow information across genes to detect and compensate the effects of unknown confounders.

Gene Suppression Technologies in High-Throughput Analysis: Front- and Back-side Applications

Our understanding of gene function and gene interactions has changed dramatically with the development of high-throughput systems. It now seems clear that any given gene interacts with a number of different partners, and in a number of different molecular pathways. Traditionally, gene function has been studied using animal knockout systems or naturally occurring mutants. RNA-based gene suppression systems for example, RNA interference or ribozymes, offer a number of advantages over the traditional systems, including ease of use, high specificity, and efficacy in nearly any biological system, and the ability to perform large-scale screens. Since their advent in the mid-1990s, DNA microarrays have been the choice for genome-wide expression analysis. The synergistic effect from the combined use of RNA-based gene suppression and molecular profiling is providing researchers with vast amounts of data. As a result, we are rapidly gaining an understanding of gene interactions and function. This review will focus primarily on gene inactivation systems that have been proven worthy of use in molecular pathway analysis when combined with microarray analysis.

Contribution of DNA and tissue microarray technology to the identification and validation of biomarkers and personalised medicine in breast cancer

Completion of the human genome project has revolutionised translational medicine. High-throughput technology now permits investigators to systematically interrogate the genome, transcriptome, proteome and metabolome. It is expected that these advances will eventually be translated into new more sensitive diagnostic tests and less toxic therapeutics. A major shift is expected in clinical oncology over the next few decades as we start to move away from currently practiced, population-based approaches to personalised medicine. In this emerging approach, the molecular and pathophysiological characteristics of an individual patient and tumour will be measured and tailored therapeutic regimens will be administered based on these profiles. One of the key steps in this process will be the identification and validation of biomarkers. Whilst great advances have been made in the discovery of putative biomarkers, disappointingly few have been translated into clinically applicable assays. It is widely believed that this is due to a lack of well-designed, thorough validation studies. Here, we review the role of DNA microarrays and tissue microarrays in the validation of biomarkers in breast cancer, with emphasis on their potential application to determine mode of personalised therapy in the future.

Application of array-based genomic and epigenomic technologies to unraveling the heterogeneous nature of breast tumors: on the road to individualized treatment

The recent application of genomic microarray technology to the molecular profiling of breast tumors has clearly demonstrated their heterogeneous nature. Targeted treatment strategies are having a clear impact on patient survival. It has also become apparent that accumulated mutations, genomic instability, epigenetic phenomena, genetic variability and environmental factors all contribute to the uniqueness of a patient's tumor. Novel genomic and epigenetic-based technologies have been or are being developed in order to greatly enhance the analysis of tumor samples including those samples previously thought unusable due to the fixation process, such as archival formalin-fixed paraffin-embedded (FFPE) samples. Patients and their tumors can now be studied with regard to genetic variation, genomic instability, gene expression, gene mutations, and methylation patterns. These areas of research are being made more accessible through genome-wide screening technologies and will, in the near future, rapidly expand our understanding of what contributes to the unique properties of each tumor and lead to the identification of genes that could be potential therapeutic targets for specific tumor subtypes. Application of these technologies to our understanding of breast cancer will undoubtedly have an impact on the individualization of treatment for breast cancer patients in the not to distant future.

Multiobjective optimization in bioinformatics and computational biology

This paper reviews the application of multiobjective optimization in the fields of bioinformatics and computational biology. A survey of existing work, organized by application area, forms the main body of the review, following an introduction to the key concepts in multiobjective optimization. An original contribution of the review is the identification of five distinct "contexts," giving rise to multiple objectives: These are used to explain the reasons behind the use of multiobjective optimization in each application area and also to point the way to potential future uses of the technique.

Molecular profiling of antipsychotic drug function: convergent mechanisms in the pathology and treatment of psychiatric disorders

Despite great progress in antipsychotic drug research, the molecular mechanisms by which these drugs work have remained elusive. High-throughput gene profiling methods have advanced this field by allowing the simultaneous investigation of hundreds to thousands of genes. However, different methodologies, choice of brain region, and drugs studied have made comparisons across different studies difficult. Because of the complexity of gene expression changes caused by drugs, teasing out the most relevant expression differences is a challenging task. One approach is to focus on gene expression changes that converge on the same systems that were previously deemed important to the pathology of psychiatric disorders. From the microarray studies performed on human postmortem brain samples from schizophrenics, the systems most implicated to be dysfunctional are synaptic machinery, oligodendrocyte/myelin function, and mitochondrial/ubiquitin metabolism. Drugs may act directly or indirectly to compensate for underlying pathological deficits in schizophrenia or via other mechanisms that converge on these pathways. Side effects, consisting of motor and metabolic dysfunction (which occur with typical and atypical drugs, respectively), also may be mediated by gene expression changes that have been reported in these studies. This article surveys both the convergent antipsychotic mechanisms and the genes that may be responsible for other effects elicited by antipsychotic drugs.

Molecular staging of prostate cancer in the year 2007

Numerous attempts towards improving patient management by molecular staging have been fruitless so far. No single molecular parameter is routinely analyzed in prostate cancer tissue. This may be partly due to genuine properties of prostate cancer that may make this tumor a difficult target. Furthermore, inherent logistical problems result in a shortage of prostate cancer tissue for research purposes. For the future, it can be hoped that the availability of more powerful molecular techniques in combination with better tissue archives will allow more rapid progress. Powerful DNA array and proteomics methods allow the systematic analysis of virtually all genes of a cancer on the DNA, RNA, and protein level. Although such approaches are sometimes labeled as "fishing expeditions," it cannot be totally disregarded that the simultaneous analysis of all genes has a high likelihood of identifying significant new information. In future, one of the major scientific challenges will be the validation of several potential biomarkers in large enough and clinically well-characterized patient cohorts. In particular, studies on needle core biopsies and hormone refractory cancers are imperatively needed for investigating the natural history of the disease or to discover potential predictive markers for radiation therapy and new therapeutic target genes to answer the clinically most important questions for optimal clinical decision making in prostate cancer patients: which patients will not require local therapy? If local therapy is needed, what is the treatment of choice? What medications should be given if metastases are present?

Treatment options and individualized medicine

Although several drug targets are identified, current strategies in therapy do not take into account that patients vary in their response to drugs, both with respect to efficacy and toxic side effects. Whereas both clinical and histopathologic predictors of prognosis are established in some diseases, a better understanding of the molecular mechanisms that determine treatment response should play an important role in the development of individualized medicine. Treatment optimization will rely on the ability to adjust treatment algorithms for use in the individual patient based on the identification and validation of the factors that critically determine treatment outcomes, including diagnosis, disease phase and characteristics, organ functions, age, and gender. Although the analysis of a single genetic marker (e.g., CYP polymorphisms) may yield significant information that predicts drug response, the prediction obtained from the analysis of several genetic and epigenetic markers is potentially more powerful in selecting patients for effective therapy, whereas sparing those who would not respond or would suffer undesirable side effects. In this chapter, several relevant examples are presented.