Extracting biology from high-dimensional biological data

Deep Learning in Mining Biological Data

Recent technological advancements in data acquisition tools allowed life scientists to acquire multimodal data from different biological application domains. Categorized in three broad types (i.e. images, signals, and sequences), these data are huge in amount and complex in nature. Mining such enormous amount of data for pattern recognition is a big challenge and requires sophisticated data-intensive machine learning techniques. Artificial neural network-based learning systems are well known for their pattern recognition capabilities, and lately their deep architectures-known as deep learning (DL)-have been successfully applied to solve many complex pattern recognition problems. To investigate how DL-especially its different architectures-has contributed and been utilized in the mining of biological data pertaining to those three types, a meta-analysis has been performed and the resulting resources have been critically analysed. Focusing on the use of DL to analyse patterns in data from diverse biological domains, this work investigates different DL architectures' applications to these data. This is followed by an exploration of available open access data sources pertaining to the three data types along with popular open-source DL tools applicable to these data. Also, comparative investigations of these tools from qualitative, quantitative, and benchmarking perspectives are provided. Finally, some open research challenges in using DL to mine biological data are outlined and a number of possible future perspectives are put forward.

Causal Queries from Observational Data in Biological Systems via Bayesian Networks: An Empirical Study in Small Networks

Biological networks are a very convenient modeling and visualization tool to discover knowledge from modern high-throughput genomics and post-genomics data sets. Indeed, biological entities are not isolated but are components of complex multilevel systems. We go one step further and advocate for the consideration of causal representations of the interactions in living systems. We present the causal formalism and bring it out in the context of biological networks, when the data is observational. We also discuss its ability to decipher the causal information flow as observed in gene expression. We also illustrate our exploration by experiments on small simulated networks as well as on a real biological data set.

Confounding Factors in the Transcriptome Analysis of an In-Vivo Exposure Experiment

Confounding factors

In transcriptomics experimentation, confounding factors frequently exist alongside the intended experimental factors and can severely influence the outcome of a transcriptome analysis. Confounding factors are regularly discussed in methodological literature, but their actual, practical impact on the outcome and interpretation of transcriptomics experiments is, to our knowledge, not documented. For instance, in-vivo experimental factors; like Individual, Sample-Composition and Time-of-Day are potentially formidable confounding factors. To study these confounding factors, we designed an extensive in-vivo transcriptome experiment (n = 264) with UVR exposure of murine skin containing six consecutive samples from each individual mouse (n = 64).

Analysis Approach

Evaluation of the confounding factors: Sample-Composition, Time-of-Day, Handling-Stress, and Individual-Mouse resulted in the identification of many genes that were affected by them. These genes sometimes showed over 30-fold expression differences. The most prominent confounding factor was Sample-Composition caused by mouse-dependent skin composition differences, sampling variation and/or influx/efflux of mobile cells. Although we can only evaluate these effects for known cell type specifically expressed genes in our complex heterogeneous samples, it is clear that the observed variations also affect the cumulative expression levels of many other non-cell-type-specific genes.

ANOVA

ANOVA analysis can only attempt to neutralize the effects of the well-defined confounding factors, such as Individual-Mouse, on the experimental factors UV-Dose and Recovery-Time. Also, by definition, ANOVA only yields reproducible gene-expression differences, but we found that these differences were very small compared to the fold changes induced by the confounding factors, questioning the biological relevance of these ANOVA-detected differences. Furthermore, it turned out that many of the differentially expressed genes found by ANOVA were also present in the gene clusters associated with the confounding factors.

Conclusion

Hence our overall conclusion is that confounding factors have a major impact on the outcome of in-vivo transcriptomics experiments. Thus the set-up, analysis, and interpretation of such experiments should be approached with the utmost prudence.

Bayesian Network Inference Enables Unbiased Phenotypic Anchoring of Transcriptomic Responses to Cigarette Smoke in Humans

Microarray-based transcriptomic analysis has been demonstrated to hold the opportunity to study the effects of human exposure to, e.g., chemical carcinogens at the whole genome level, thus yielding broad-ranging molecular information on possible carcinogenic effects. Since genes do not operate individually but rather through concerted interactions, analyzing and visualizing networks of genes should provide important mechanistic information, especially upon connecting them to functional parameters, such as those derived from measurements of biomarkers for exposure and carcinogenic risk. Conventional methods such as hierarchical clustering and correlation analyses are frequently used to address these complex interactions but are limited as they do not provide directional causal dependence relationships. Therefore, our aim was to apply Bayesian network inference with the purpose of phenotypic anchoring of modified gene expressions. We investigated a use case on transcriptomic responses to cigarette smoking in humans, in association with plasma cotinine levels as biomarkers of exposure and aromatic DNA-adducts in blood cells as biomarkers of carcinogenic risk. Many of the genes that appear in the Bayesian networks surrounding plasma cotinine, and to a lesser extent around aromatic DNA-adducts, hold biologically relevant functions in inducing severe adverse effects of smoking. In conclusion, this study shows that Bayesian network inference enables unbiased phenotypic anchoring of transcriptomics responses. Furthermore, in all inferred Bayesian networks several dependencies are found which point to known but also to new relationships between the expression of specific genes, cigarette smoke exposure, DNA damaging-effects, and smoking-related diseases, in particular associated with apoptosis, DNA repair, and tumor suppression, as well as with autoimmunity.

Unraveling the sperm proteome and post-genomic pathways associated with sperm nuclear DNA fragmentation

PURPOSE

Sperm DNA fragmentation has been suggested as a marker for infertility diagnosis and prognosis. Hence, understanding its impact on male physiology and post-genomic pathways would be clinically important. We performed the proteomics and functional enrichment analyses of viable spermatozoa from ejaculates with low and high sperm DNA fragmentation to identify protein expression and pathways altered in association with sperm DNA fragmentation.

METHODS

Sperm DNA fragmentation using the Comet assay and the Komet 6.0.1 software was assessed in raw samples from 89 subjects from a human reproduction service. The Low and High sperm DNA fragmentation groups were formed according to the Olive Tail Moment variable. Spermatozoa proteins from these groups were pooled and analyzed by a shotgun proteomic approach (2D nanoUPLC-ESI-MS(E)). Differentially expressed proteins were used for a functional enrichment study.

RESULTS

Two hundred and fifty-seven proteins were identified or quantified in sperm from the Low and High sperm DNA fragmentation groups. Of these, seventy-one proteins were exclusively or overexpressed in the Low group, whereas twenty-three proteins were exclusively or overexpressed in the High group. One hundred and sixty-three proteins were conserved between these groups. We also functionally related the differentially expressed proteins in viable spermatozoa from the groups. Processes such as triacylglycerol metabolism, energy production, protein folding, response to unfolded proteins, and cellular detoxification were found to be altered in these cells.

CONCLUSIONS

Sperm DNA fragmentation is associated with differential protein expression in viable spermatozoa. These proteins may potentially be used as biomarkers for sperm DNA integrity.

Sperm nuclear DNA fragmentation rate is associated with differential protein expression and enriched functions in human seminal plasma

OBJECTIVE

To analyse the proteomic profile of seminal plasma with the aim of identifying the proteins and post-genomic pathways associated with sperm DNA fragmentation.

MATERIALS AND METHODS

A cross-sectional study including 89 subjects from a human reproduction service was carried out. All semen samples were assessed for sperm DNA fragmentation using a comet assay. Results from 60 sperm were analysed using Komet 6.0.1 software and the 'Olive tail moment' variable was used to stratify these into low and high sperm DNA fragmentation groups. Seminal plasma proteins from the two groups were pooled and used for proteomic analysis. Quantitative data were used for functional enrichment studies.

RESULTS

Seventy-two proteins were identified or quantified in seminal plasma. Of these, nine were differentially expressed in the low group and 21 in the high group. Forty-two proteins were conserved between these groups. Functional enrichment analysis indicated that sperm DNA fragmentation was related to functions such as lipoprotein particle remodelling and regulation, fatty acid binding and immune response. Proteins found exclusively in the low group may be involved in correcting spermatogenesis and/or improving sperm function. Proteins in the high group were associated with increased innate immune response, sperm motility and/or maturation and inhibition of mitochondrial apoptosis.

CONCLUSION

Protein expression and post-genomic pathways of seminal plasma differ according to the rate of sperm DNA integrity.

Cell-type-specific predictive network yields novel insights into mouse embryonic stem cell self-renewal and cell fate

Self-renewal, the ability of a stem cell to divide repeatedly while maintaining an undifferentiated state, is a defining characteristic of all stem cells. Here, we clarify the molecular foundations of mouse embryonic stem cell (mESC) self-renewal by applying a proven Bayesian network machine learning approach to integrate high-throughput data for protein function discovery. By focusing on a single stem-cell system, at a specific developmental stage, within the context of well-defined biological processes known to be active in that cell type, we produce a consensus predictive network that reflects biological reality more closely than those made by prior efforts using more generalized, context-independent methods. In addition, we show how machine learning efforts may be misled if the tissue specific role of mammalian proteins is not defined in the training set and circumscribed in the evidential data. For this study, we assembled an extensive compendium of mESC data: ∼2.2 million data points, collected from 60 different studies, under 992 conditions. We then integrated these data into a consensus mESC functional relationship network focused on biological processes associated with embryonic stem cell self-renewal and cell fate determination. Computational evaluations, literature validation, and analyses of predicted functional linkages show that our results are highly accurate and biologically relevant. Our mESC network predicts many novel players involved in self-renewal and serves as the foundation for future pluripotent stem cell studies. This network can be used by stem cell researchers (at http://StemSight.org) to explore hypotheses about gene function in the context of self-renewal and to prioritize genes of interest for experimental validation.

Systems biology in animal sciences

Systems biology is a rapidly expanding field of research and is applied in a number of biological disciplines. In animal sciences, omics approaches are increasingly used, yielding vast amounts of data, but systems biology approaches to extract understanding from these data of biological processes and animal traits are not yet frequently used. This paper aims to explain what systems biology is and which areas of animal sciences could benefit from systems biology approaches. Systems biology aims to understand whole biological systems working as a unit, rather than investigating their individual components. Therefore, systems biology can be considered a holistic approach, as opposed to reductionism. The recently developed 'omics' technologies enable biological sciences to characterize the molecular components of life with ever increasing speed, yielding vast amounts of data. However, biological functions do not follow from the simple addition of the properties of system components, but rather arise from the dynamic interactions of these components. Systems biology combines statistics, bioinformatics and mathematical modeling to integrate and analyze large amounts of data in order to extract a better understanding of the biology from these huge data sets and to predict the behavior of biological systems. A 'system' approach and mathematical modeling in biological sciences are not new in itself, as they were used in biochemistry, physiology and genetics long before the name systems biology was coined. However, the present combination of mass biological data and of computational and modeling tools is unprecedented and truly represents a major paradigm shift in biology. Significant advances have been made using systems biology approaches, especially in the field of bacterial and eukaryotic cells and in human medicine. Similarly, progress is being made with 'system approaches' in animal sciences, providing exciting opportunities to predict and modulate animal traits.

Graphs and networks in chemical and biological informatics: past, present and future

Chemical and biological network analysis has recently garnered intense interest from the perspective of drug design and discovery. While graph theoretic concepts have a long history in chemistry – predating quantum mechanics – and graphical measures of chemical structures date back to the 1970s, it is only recently with the advent of public repositories of information and availability of high-throughput assays and computational resources that network analysis of large-scale chemical networks, such as protein–protein interaction networks, has become possible. Drug design and discovery are undergoing a paradigm shift, from the notion of ‘one target, one drug’ to a much more nuanced view that relies on multiple sources of information: genomic, proteomic, metabolomic and so on. This holistic view of drug design is an incredibly daunting undertaking still very much in its infancy. Here, we focus on current developments in graph- and network-centric approaches in chemical and biological informatics, with particular reference to applications in the fields of SAR modeling and drug design. Key insights from the past suggest a path forward via visualization and fusion of multiple sources of chemical network data.

A systems biology approach for detecting toxicity-related hotspots inside protein interaction networks

Drug-induced neutropenia can be fatal when severe and therefore requires an improved understanding of its mechanism(s) of toxicity. Systems biology provides an opportunity to understand adverse events after drug administration using analysis of biomolecular networks. In this study, a human protein interaction network was analyzed to identify proteins that are most central to topological paths connecting a drug's target proteins to hematopoiesis-related proteins. For a set of non-immune neutropenia inducing drugs, 9 proteins were found to be common to putative signaling paths across all drugs evaluated. All 9 proteins showed relevance to neutrophil biology. Geneset enrichment analysis showed that proteins associated with cancer-related processes such as apoptosis provide topological linkages between drug targets and proteins involved in neutrophil production. The algorithm can be applied towards analysis of any toxicity where the drugs and the physiological processes involved in the toxic mechanism are known.

Recent advances of metabolomics in plant biotechnology

Biotechnology, including genetic modification, is a very important approach to regulate the production of particular metabolites in plants to improve their adaptation to environmental stress, to improve food quality, and to increase crop yield. Unfortunately, these approaches do not necessarily lead to the expected results due to the highly complex mechanisms underlying metabolic regulation in plants. In this context, metabolomics plays a key role in plant molecular biotechnology, where plant cells are modified by the expression of engineered genes, because we can obtain information on the metabolic status of cells via a snapshot of their metabolome. Although metabolome analysis could be used to evaluate the effect of foreign genes and understand the metabolic state of cells, there is no single analytical method for metabolomics because of the wide range of chemicals synthesized in plants. Here, we describe the basic analytical advancements in plant metabolomics and bioinformatics and the application of metabolomics to the biological study of plants.

Confidence from uncertainty--a multi-target drug screening method from robust control theory

Background

Robustness is a recognized feature of biological systems that evolved as a defence to environmental variability. Complex diseases such as diabetes, cancer, bacterial and viral infections, exploit the same mechanisms that allow for robust behaviour in healthy conditions to ensure their own continuance. Single drug therapies, while generally potent regulators of their specific protein/gene targets, often fail to counter the robustness of the disease in question. Multi-drug therapies offer a powerful means to restore disrupted biological networks, by targeting the subsystem of interest while preventing the diseased network from reconciling through available, redundant mechanisms. Modelling techniques are needed to manage the high number of combinatorial possibilities arising in multi-drug therapeutic design, and identify synergistic targets that are robust to system uncertainty.

Results

We present the application of a method from robust control theory, Structured Singular Value or μ- analysis, to identify highly effective multi-drug therapies by using robustness in the face of uncertainty as a new means of target discrimination. We illustrate the method by means of a case study of a negative feedback network motif subject to parametric uncertainty.

Conclusions

The paper contributes to the development of effective methods for drug screening in the context of network modelling affected by parametric uncertainty. The results have wide applicability for the analysis of different sources of uncertainty like noise experienced in the data, neglected dynamics, or intrinsic biological variability.

Building promoter aware transcriptional regulatory networks using siRNA perturbation and deepCAGE

Perturbation and time-course data sets, in combination with computational approaches, can be used to infer transcriptional regulatory networks which ultimately govern the developmental pathways and responses of cells. Here, we individually knocked down the four transcription factors PU.1, IRF8, MYB and SP1 in the human monocyte leukemia THP-1 cell line and profiled the genome-wide transcriptional response of individual transcription starting sites using deep sequencing based Cap Analysis of Gene Expression. From the proximal promoter regions of the responding transcription starting sites, we derived de novo binding-site motifs, characterized their biological function and constructed a network. We found a previously described composite motif for PU.1 and IRF8 that explains the overlapping set of transcriptional responses upon knockdown of either factor.

Thermus thermophilus proteins that are differentially expressed in response to growth temperature and their implication in thermoadaptation

As a kind of important extremophiles to realize the adaptation of life at high temperatures, thermophiles have attracted extensive studies. However, the pathways of thermophile proteins related to thermoadaptation remain to be addressed. Our study showed that there existed two types of protein profiles for the thermophile Thermus thermophilus wl in response to temperature change. One of them came from cultures growing below 65 degrees C, which was close to the optimal growth temperature, and another from cultures at or above 65 degrees C. These protein profiles were confirmed by Northern blots. On the basis of the proteomic and computational analyses, it was found that the thermophile proteins related to thermoadaptation might be involved in metabolic pathways as well as the stabilities and modifications of DNA and proteins. Interestingly, for the basic metabolism glycolysis, the phosphoglucomutase was up-regulated at below-optimum temperature, while the glyceraldehyde-3-phosphate dehydrogenase was up-regulated at above-optimum temperature, suggesting that different regulations might be used for basic metabolism at different temperatures. To characterize the proteins in response to high temperatures, superoxide dismutase (SOD), an important enzyme in organism to remove free radical produced in stress environment such as high temperature, was selected as a target protein for this investigation. SOD was inactivated to construct a SOD mutant. The results showed that the SOD protein was essential in thermoadaptation of T. thermophilus. Our study, therefore, presented the thermophile proteins required for thermoadaptation and their possible pathways in thermoadaptation.

Approaches to advancing quantitative human health risk assessment of environmental chemicals in the post-genomic era

The contribution of genomics and associated technologies to human health risk assessment for environmental chemicals has focused largely on elucidating mechanisms of toxicity, as discussed in other articles in this issue. However, there is interest in moving beyond hazard characterization to making more direct impacts on quantitative risk assessment (QRA)--i.e., the determination of toxicity values for setting exposure standards and cleanup values. We propose that the evolution of QRA of environmental chemicals in the post-genomic era will involve three, somewhat overlapping phases in which different types of approaches begin to mature. The initial focus (in Phase I) has been and continues to be on "augmentation" of weight of evidence--using genomic and related technologies qualitatively to increase the confidence in and scientific basis of the results of QRA. Efforts aimed towards "integration" of these data with traditional animal-based approaches, in particular quantitative predictors, or surrogates, for the in vivo toxicity data to which they have been anchored are just beginning to be explored now (in Phase II). In parallel, there is a recognized need for "expansion" of the use of established biomarkers of susceptibility or risk of human diseases and disorders for QRA, particularly for addressing the issues of cumulative assessment and population risk. Ultimately (in Phase III), substantial further advances could be realized by the development of novel molecular and pathway-based biomarkers and statistical and in silico models that build on anticipated progress in understanding the pathways of human diseases and disorders. Such efforts would facilitate a gradual "reorientation" of QRA towards approaches that more directly link environmental exposures to human outcomes.

Dynamical approaches to modeling developmental gene regulatory networks

The network of interacting regulatory signals within a cell comprises one of the most complex and powerful computational systems in biology. Gene regulatory networks (GRNs) play a key role in transforming the information encoded in a genome into morphological form. To achieve this feat, GRNs must respond to and integrate environmental signals with their internal dynamics in a robust and coordinated fashion. The highly dynamic nature of this process lends itself to interpretation and analysis in the language of dynamical models. Modeling provides a means of systematically untangling the complicated structure of GRNs, a framework within which to simulate the behavior of reconstructed systems and, in some cases, suites of analytic tools for exploring that behavior and its implications. This review provides a general background to the idea of treating a regulatory network as a dynamical system, and describes a variety of different approaches that have been taken to the dynamical modeling of GRNs.

From transcriptome analysis to immunogenomics: current status and future direction

In 1994, we pioneered a complementary DNA (cDNA) sequencing project that aimed to predict the primary structures of unknown human proteins. Although our cDNA project was focused on the sequencing of large cDNAs, the following cDNA sequencing projects conducted by other groups have more extensively characterized mammalian transcriptome. In parallel, many groups have made a tremendous amount of effort to develop various resources for functional human genomics. In this context, to demonstrate the power of functional genomic approaches in practice, we have applied them for a comprehensive understanding of the immune system, which we term 'immunogenomics'. This mini-review first describes the historical background of our cDNA project and then provides perspectives on the present and future of immunogenomics based on our experiences.

Genome-wide gene expression profiling and mutation analysis of Saudi patients with Canavan disease

PURPOSE

Canavan disease, caused by a deficiency of aspartoacylase, is one of the most common cerebral degenerative diseases of infancy. The aims of this study were to identify the mutations associated with Canavan disease in Saudi Arabia and to identify differentially expressed genes likely to contribute to the development of this disease.

METHODS

Polymerase chain reaction, long polymerase chain reaction, multiplex ligation-dependent probe amplification, sequencing, array comparative genomic hybridization (aCGH), and global gene expression profiling were used to determine putative mutations and likely gene signatures in cultured fibroblasts of patients from Saudi Arabia.

RESULTS

One novel and one known large deletion and two previously known mutations (IVS4 + 1G>T and G27R) were identified. Compared with controls, 1440 genes were significantly modulated in Canavan patients (absolute fold change [FC] > or =4). Genome-wide gene expression profiling results indicated that some genes, involved in apoptosis, muscle contraction and development, mitochondrial oxidation, inflammation and glutamate, and aspartate metabolism, were significantly dysregulated.

CONCLUSIONS

Our findings indicate that the presence of muscle weakness and hypotonia in patients may be associated with the dysregulated gene activities of cell motility, muscle contraction and development, actin binding, and cytoskeletal-related activities. Overall, these observations are in accordance with previous studies performed in a knockout mouse model.

HMDB: a knowledgebase for the human metabolome

The Human Metabolome Database (HMDB, http://www.hmdb.ca) is a richly annotated resource that is designed to address the broad needs of biochemists, clinical chemists, physicians, medical geneticists, nutritionists and members of the metabolomics community. Since its first release in 2007, the HMDB has been used to facilitate the research for nearly 100 published studies in metabolomics, clinical biochemistry and systems biology. The most recent release of HMDB (version 2.0) has been significantly expanded and enhanced over the previous release (version 1.0). In particular, the number of fully annotated metabolite entries has grown from 2180 to more than 6800 (a 300% increase), while the number of metabolites with biofluid or tissue concentration data has grown by a factor of five (from 883 to 4413). Similarly, the number of purified compounds with reference to NMR, LC-MS and GC-MS spectra has more than doubled (from 380 to more than 790 compounds). In addition to this significant expansion in database size, many new database searching tools and new data content has been added or enhanced. These include better algorithms for spectral searching and matching, more powerful chemical substructure searches, faster text searching software, as well as dedicated pathway searching tools and customized, clickable metabolic maps. Changes to the user-interface have also been implemented to accommodate future expansion and to make database navigation much easier. These improvements should make the HMDB much more useful to a much wider community of users.

Combined line-cross and half-sib QTL analysis in Duroc-Pietrain population

A Duroc-Pietrain resource population was built to detect quantitative trait loci (QTL) that affect growth, carcass composition, and pork quality. The data were analyzed by applying three least-squares Mendelian models: a line-cross (LC) model, a half-sib (HS) model, and a combined LC and HS model (CB), which enabled the detection of QTL that had fixed, equal, and different allele frequencies for alternate breed alleles, respectively. Permutation tests were performed to determine 5% chromosome-wide and 5% genome-wide threshold values. A total of 40 (137) QTL were detected at the 5% genome-wide (chromosome-wide) level for the 35 traits analyzed. Of the 137 QTL detected, 62 were classified as the LC type (LC-QTL), 47 as the HS type (HS-QTL), and 28 as the CB type (CB-QTL). The results indicate that implementation of a series of model-based framework is not only beneficial to detect QTL, but also provides us with a new and more robust interpretation from which further methodology could be developed.

Gene-expression profiling in epithelial ovarian cancer

DNA-microarray technology has made it possible to simultaneously analyze the expression of thousands of genes in a small sample of tumor tissue. In epithelial ovarian cancer, gene-expression profiling has been used to provide prognostic information, to predict response to first-line platinum-based chemotherapy, and to discriminate between different histologic subtypes. Furthermore, DNA-microarray technology might permit identification of novel markers for early detection of disease and provide insights into the mechanisms of cancer growth and chemotherapy resistance. In this Review, we summarize the contributions of gene-expression profiling to the diagnosis and management of epithelial ovarian cancer and discuss ways in which this technique could become a useful tool in clinical management.

Towards a systems level analysis of health and nutrition

Although theoretical systems analysis has been available for over half a century, the recent advent of omic high-throughput analytical platforms along with the integration of individual tools and technologies has given rise to the field of modern systems biology. Coupled with information technology, bioinformatics, knowledge management and powerful mathematical models, systems biology has opened up new vistas in our understanding of complex biological systems. Currently there are two distinct approaches that include the inductively driven computational systems biology (bottom-up approach) and the deductive data-driven top-down analysis. Such approaches offer enormous potential in the elucidation of disease as well as defining key pathways and networks involved in optimal human health and nutrition. The tools and technologies now available in systems biology analyses offer exciting opportunities to develop the emerging areas of personalized medicine and individual nutritional profiling.

Gene expression microarray data analysis demystified

The increasing use of gene expression microarrays, and depositing of the resulting data into public repositories, means that more investigators are interested in using the technology either directly or through meta analysis of the publicly available data. The tools available for data analysis have generally been developed for use by experts in the field, making them difficult to use by the general research community. For those interested in entering the field, especially those without a background in statistics, it is difficult to understand why experimental results can be so variable. The purpose of this review is to go through the workflow of a typical microarray experiment, to show that decisions made at each step, from choice of platform through statistical analysis methods to biological interpretation, are all sources of this variability.

The multidisciplinary management of gastrointestinal cancer. The use of molecular markers in the diagnosis and treatment of colorectal cancer

Rapidly growing insights into the molecular biology of colorectal cancer led to high hopes for the identification of molecular markers to be used in optimised and tailored treatment regimens for this disorder. However, no molecular marker has yet made it into daily practice. In this review we will discuss some of the potential molecular markers, focus on the lessons learnt from marker development and identify strategies for the future.