Search

Reconstructing a gene network from high-throughput molecular data is an important but challenging task, as the number of parameters to estimate easily is much larger than the sample size. A conventional remedy is to regularize or penalize the model likelihood. In network models, this is often done locally in the neighborhood of each node or gene. However, estimation of the many regularization parameters is often difficult and can result in large statistical uncertainties. In this paper we propose to combine local regularization with global shrinkage of the regularization parameters to borrow strength between genes and improve inference. We employ a simple Bayesian model with nonsparse, conjugate priors to facilitate the use of fast variational approximations to posteriors. We discuss empirical Bayes estimation of hyperparameters of the priors, and propose a novel approach to rank-based posterior thresholding. Using extensive model- and data-based simulations, we demonstrate that the proposed inference strategy outperforms popular (sparse) methods, yields more stable edges, and is more reproducible. The proposed method, termed ShrinkNet, is then applied to Glioblastoma to investigate the interactions between genes associated with patient survival.

Background: Combining clinical and molecular data types may potentially improve prediction accuracy of a classifier. However, currently there is a shortage of effective and efficient statistical and bioinformatic tools for true integrative data analysis. Existing integrative classifiers have two main disadvantages: First, coarse combination may lead to subtle contributions of one data type to be overshadowed by more obvious contributions of the other. Second, the need to measure both data types for all patients may be both unpractical and (cost) inefficient.Results: We introduce a novel classification method, a stepwise classifier, which takes advantage of the distinct classification power of clinical data and high-dimensional molecular data. We apply classification algorithms to two data types independently, starting with the traditional clinical risk factors. We only turn to relatively expensive molecular data when the uncertainty of prediction result from clinical data exceeds a predefined limit. Experimental results show that our approach is adaptive: the proportion of samples that needs to be re-classified using molecular data depends on how much we expect the predictive accuracy to increase when re-classifying those samples.Conclusions: Our method renders a more cost-efficient classifier that is at least as good, and sometimes better, than one based on clinical or molecular data alone. Hence our approach is not just a classifier that minimizes a particular loss function. Instead, it aims to be cost-efficient by avoiding molecular tests for a potentially large subgroup of individuals; moreover, for these individuals a test result would be quickly available, which may lead to reduced waiting times (for diagnosis) and hence lower the patients distress. Stepwise classification is implemented in R-package stepwiseCM and available at the Bioconductor website. © 2011 Obulkasim et al; licensee BioMed Central Ltd.

Background: Determining a suitable sample size is an important step in the planning of microarray experiments. Increasing the number of arrays gives more statistical power, but adds to the total cost of the experiment. Several approaches for sample size determination have been developed for expression array studies, but so far none has been proposed for array comparative genomic hybridization (aCGH).Results: Here we explore power calculations for aCGH experiments comparing two groups. In a pilot experiment CGHpower estimates the biological diversity between groups and provides a statistical framework for estimating average power as a function of sample size. As the method requires pilot data, it can be used either in the planning stage of larger studies or in estimating the power achieved in past experiments.Conclusions: The proposed method relies on certain assumptions. According to our evaluation with public and simulated data sets, they do not always hold true. Violation of the assumptions typically leads to unreliable sample size estimates. Despite its limitations, this method is, at least to our knowledge, the only one currently available for performing sample size calculations in the context of aCGH. Moreover, the implementation of the method provides diagnostic plots that allow critical assessment of the assumptions on which it is based and hence on the feasibility and reliability of the sample size calculations in each case.The CGHpower web application and the program outputs from evaluation data sets can be freely accessed at http://www.cangem.org/cghpower/. © 2010 Scheinin et al; licensee BioMed Central Ltd.

Background: Ratio-based analysis is the current standard for the analysis of dual-color microarray data. Indeed, this method provides a powerful means to account for potential technical variations such as differences in background signal, spot size and spot concentration. However, current high density dual-color array platforms are of very high quality, and inter-array variance has become much less pronounced. We therefore raised the question whether it is feasible to use an intensity-based analysis rather than ratio-based analysis of dual-color microarray datasets. Furthermore, we compared performance of both ratio- and intensity-based analyses in terms of reproducibility and sensitivity for differential gene expression.Results: By analyzing three distinct and technically replicated datasets with either ratio- or intensity-based models, we determined that, when applied to the same dataset, intensity-based analysis of dual-color gene expression experiments yields 1) more reproducible results, and 2) is more sensitive in the detection of differentially expressed genes. These effects were most pronounced in experiments with large biological variation and complex hybridization designs. Furthermore, a power analysis revealed that for direct two-group comparisons above a certain sample size, ratio-based models have higher power, although the difference with intensity-based models is very small.Conclusions: Intensity-based analysis of dual-color datasets results in more reproducible results and increased sensitivity in the detection of differential gene expression than the analysis of the same dataset with ratio-based analysis. Complex dual-color setups such as interwoven loop designs benefit most from ignoring the array factor. The applicability of our approach to array platforms other than dual-color needs to be further investigated. © 2010 Bossers et al; licensee BioMed Central Ltd.

Background: The active form of Vitamin D, 1,25-dihydroxyvitamin D

Background. Gastric cancers frequently show chromosomal alterations which can cause activation of oncogenes, and/or inactivation of tumour suppressor genes. In gastric cancer several chromosomal regions are described to be frequently lost, but for most of the regions, no tumour suppressor genes have been identified yet. The present study aimed to identify tumour suppressor genes inactivated by nonsense mutation and deletion in gastric cancer by means of GINI (gene identification by nonsense mediated decay inhibition) and whole genome copy number analysis. Methods. Two non-commercial gastric cancer cell lines, GP202 and IPA220, were transfected with siRNA directed against UPF1, to specifically inhibit the nonsense mediated decay (NMD) pathway, and with siRNA directed against non-specific siRNA duplexes (CVII) as a control. Microarray expression experiments were performed in triplicate on 4 × 44 K Agilent arrays by hybridizing RNA from UPF1-transfected cells against non-specific CVII-transfected cells. In addition, array CGH of the two cell lines was performed on 4 × 44K agilent arrays to obtain the DNA copy number profiles. Mutation analysis of GINI candidates was performed by sequencing. Results. UPF1 expression was reduced for ≥70% and ≥80% in the GP202 and IPA220 gastric cancer cell lines, respectively. Integration of array CGH and microarray expression data provided a list of 134 and 50 candidate genes inactivated by nonsense mutation and deletion for GP202 and IPA220, respectively. We selected 12 candidate genes for mutation analysis. Of these, sequence analysis was performed on 11 genes. One gene, PLA2G4A, showed a silent mutation, and in two genes, CTSA and PTPRJ, missense mutations were detected. No nonsense mutations were detected in any of the 11 genes tested. Conclusion. Although UPF1 was substantially repressed, thus resulting in the inhibition of the NMD system, we did not find genes inactivated by nonsense mutations. Our results show that the GINI strat

Objective: This study aimed to identify the oncogenes at 20q involved in colorectal adenoma to carcinoma progression by measuring the effect of 20q gain on mRNA expression of genes in this amplicon. Methods: Segmentation of DNA copy number changes on 20q was performed by array CGH (comparative genomic hybridisation) in 34 non-progressed colorectal adenomas, 41 progressed adenomas (ie, adenomas that present a focus of cancer) and 33 adenocarcinomas. Moreover, a robust analysis of altered expression of genes in these segments was performed by microarray analysis in 37 adenomas and 31 adenocarcinomas. Protein expression was evaluated by immunohistochemistry on tissue microarrays. Results: The genes C20orf24, AURKA, RNPC1, TH1L, ADRM1, C20orf20 and TCFL5, mapping at 20q, were significantly overexpressed in carcinomas compared with adenomas as a consequence of copy number gain of 20q. Conclusion: This approach revealed C20orf24, AURKA, RNPC1, TH1L, ADRM1, C20orf20 and TCFL5 genes to be important in chromosomal instability-related adenoma to carcinoma progression. These genes therefore may serve as highly specific biomarkers for colorectal cancer with potential clinical applications.

Background:MicroRNAs are small non-coding RNA molecules, which regulate central mechanisms of tumorigenesis. In colorectal tumours, the combination of gain of 8q and 13q is one of the major factors associated with colorectal adenoma to adenocarcinoma progression. Functional studies on the miR-17-92 cluster localised on 13q31 have shown that its transcription is activated by c-myc, located on 8q, and that it has oncogenic activities. We investigated the contribution of the miR-17-92 cluster during colorectal adenoma to adenocarcinoma progression.Methods:Expression levels of the miR-17-92 cluster were determined in 55 colorectal tumours and in 10 controls by real-time RT-PCR. Messenger RNA c-myc expression was also determined by real-time RT-PCR in 48 tumours with array comparative genomic hybridisation (aCGH) data available.Results:From the six members of the miR-17-92 cluster, all except miR-18a, showed significant increased expression in colorectal tumours with miR-17-92 locus gain compared with tumours without miR-17-92 locus gain. Unsupervised cluster analysis clustered the tumours based on the presence of miR-17-92 locus gain. Significant correlation between the expression of c-myc and the six miRNAs was also found.Conclusion:Increased expression of miR-17-92 cluster during colorectal adenoma to adenocarcinoma progression is associated to DNA copy number gain of miR17-92 locus on 13q31 and c-myc expression. © 2009 Cancer Research UK.

Background. It is well known that a persistent infection with high-risk human papillomavirus (hrHPV) is causally involved in the development of squamous cell carcinomas of the uterine cervix (CxSCCs) and a subset of SCCs of the head and neck (HNSCCs). The latter differ from hrHPV-negative HNSCCs at the clinical and molecular level. Methods. To determine whether hrHPV-associated SCCs arising from different organs have specific chromosomal alterations in common, we compared genome-wide chromosomal profiles of 10 CxSCCs (all hrHPV-positive) with 12 hrHPV-positive HNSCCs and 30 hrHPV-negative HNSCCs. Potential organ-specific alterations and alterations shared by SCCs in general were investigated as well. Results. Unsupervised hierarchical clustering resulted in one mainly hrHPV-positive and one mainly hrHPV-negative cluster. Interestingly, loss at 13q and gain at 20q were frequent in HPV-positive carcinomas of both origins, but uncommon in hrHPV-negative HNSCCs, indicating that these alterations are associated with hrHPV-mediated carcinogenesis. Within the group of hrHPV-positive carcinomas, HNSCCs more frequently showed gains of multiple regions at 8q whereas CxSCCs more often showed loss at 17p. Finally, gains at 3q24-29 and losses at 11q22.3-25 were frequent (>50%) in all sample groups. Conclusion. In this study hrHPV-specific, organ-specific, and pan-SCC chromosomal alterations were identified. The existence of hrHPV-specific alterations in SCCs of different anatomical origin, suggests that these alterations are crucial for hrHPV-mediated carcinogenesis. © 2009 Wilting et al; licensee BioMed Central Ltd.

Background: We consider effects of dependence among variables of high-dimensional data in multiple hypothesis testing problems, in particular the False Discovery Rate (FDR) control procedures. Recent simulation studies consider only simple correlation structures among variables, which is hardly inspired by real data features. Our aim is to systematically study effects of several network features like sparsity and correlation strength by imposing dependence structures among variables using random correlation matrices. Results: We study the robustness against dependence of several FDR procedures that are popular in microarray studies, such as Benjamin-Hochberg FDR, Storey's q-value, SAM and resampling based FDR procedures. False Non-discovery Rates and estimates of the number of null hypotheses are computed from those methods and compared. Our simulation study shows that methods such as SAM and the q-value do not adequately control the FDR to the level claimed under dependence conditions. On the other hand, the adaptive Benjamini-Hochberg procedure seems to be most robust while remaining conservative. Finally, the estimates of the number of true null hypotheses under various dependence conditions are variable. Conclusion: We discuss a new method for efficient guided simulation of dependent data, which satisfy imposed network constraints as conditional independence structures. Our simulation set-up allows for a structural study of the effect of dependencies on multiple testing criterions and is useful for testing a potentially new method on π

Background: Oligoarrays have become an accessible technique for exploring the transcriptome, but it is presently unclear how absolute transcript data from this technique compare to the data achieved with tag-based quantitative techniques, such as massively parallel signature sequencing (MPSS) and serial analysis of gene expression (SAGE). By use of the TransCount method we calculated absolute transcript concentrations from spotted oligoarray intensities, enabling direct comparisons with tag counts obtained with MPSS and SAGE. The tag counts were converted to number of transcripts per cell by assuming that the sum of all transcripts in a single cell was 5·10

Summary: CGHcall achieves high calling accuracy for array CGH data by effective use of breakpoint information from segmentation and by inclusion of several biological concepts that are ignored by existing algorithms. The algorithm is validated for simulated and verified real array CGH data. By incorporating more than three classes, CGHcall improves detection of single copy gains and amplifications. Moreover, it allows effective inclusion of chromosome arm information. © The Author 2007. Published by Oxford University Press. All rights reserved.

A method providing absolute transcript concentrations from spotted microarray intensity data is presented. Number of transcripts per µg total RNA, mRNA or per cell, are obtained for each gene, enabling comparisons of transcript levels within and between tissues. The method is based on Bayesian statistical modelling incorporating available information about the experiment from target preparation to image analysis, leading to realistically large confidence intervals for estimated concentrations. The method was validated in experiments using transcripts at known concentrations, showing accuracy and reproducibility of estimated concentrations, which were also in excellent agreement with results from quantitative real-time PCR. We determined the concentration for 10 157 genes in cervix cancers and a pool of cancer cell lines and found values in the range of 105–1010 transcripts per µg total RNA. The precision of our estimates was sufficiently high to detect significant concentration differences between two tumours and between different genes within the same tumour, comparisons that are not possible with standard intensity ratios. Our method can be used to explore the regulation of pathways and to develop individualized therapies, based on absolute transcript concentrations. It can be applied broadly, facilitating the construction of the transcriptome, continuously updating it by integrating future data.

Summary: We compute P-values, based on the Wilcoxon test with ties, to compare two conditions with array comparative genomic hybridization data, and we provide a simple interface to export and plot these P-values.

Gemcitabine is a commonly used therapy for many solid tumors. Acquired resistance to this nucleoside analogue, however, diminishes the long-term effectiveness in a majority of patients. To better define the molecular background of gemcitabine resistance, a mouse colon tumor was selected during successive rounds of transplantation with continued treatment of gemcitabine. Expression microarray analysis was applied to determine which genes are consistently and highly overexpressed or underexpressed in the resistant versus the nonresistant tumor. For the statistical interpretation of the microarray data, a parametric model was implemented, which returns model-based differential gene expression (log-) ratios and their uncertainties. This defined a set of 13 genes, putatively responsible for the gemcitabine resistance in solid tumors. One of these, RRM1, was previously identified as an important marker for gemcitabine resistance in human cell lines. Five of the 13 genes, including RRM1, are located within a 3 Mb region at chromosome 7E1 of which four are highly overexpressed, suggesting a chromosomal amplification. Therefore, chromosomal copy number changes were measured, using oligo array comparative genomic hybridization. A narrow and high amplification area was identified on 7E1 that encompassed all five genes. In addition, reduced RNA expression of two other genes at 8E1 encoding COX4I1 and RPL13 could be explained by a decrease in chromosomal copy number on chromosome 8. In conclusion, the array comparative genomic hybridization biologically validates our statistical approach and shows that gemcitabine is capable to select for chromosomally aberrant tumor cells, where changed gene expression levels lead to drug resistance.

We review several commonly used methods for the design and analysis of microarray data. To begin with, some experimental design issues are addressed. Several approaches for pre‐processing the data (filtering and normalization) before the statistical analysis stage are then discussed. A common first step in this type of analysis is gene selection based on statistical testing. Two approaches, permutation and model‐based methods are explained and we emphasize the need to correct for multiple testing. Moreover, powerful approaches based on gene sets are mentioned. Clustering of either genes or samples is frequently performed when analyzing microarray data. We summarize the basics of both supervised and unsupervised clustering (classification). The latter may be of use for creating diagnostic arrays, for example. Construction of biological networks, such as pathways, is a statistically challenging but complex task that is a relatively new development and hence mentioned only briefly. We finish with some remarks on literature and software. The emphasis in this paper is on the philosophy behind several statistical issues and on a critical interpretation of microarray related analysis methods