Primary

Hierarchical clustering is a common procedure for identifying structure in a data set, and this is frequently used for organizing genomic data. Although more advanced clustering algorithms are available, the simplicity and visual appeal of hierarchical clustering has made it ubiquitous in gene expression data analysis. Hence, even minor improvements in this framework would have significant impact. There is currently no simple and systematic way of assessing and displaying the significance of various clusters in a resulting dendrogram without making certain distributional assumptions or ignoring gene-specific variances. In this work, we introduce a permutation test based on comparing the within-cluster structure of the observed data with those of sample datasets obtained by permuting the cluster membership. We carry out this test at each node of the dendrogram using a statistic derived from the singular value decomposition of variance matrices. The p-values thus obtained provide insight into the significance of each cluster division. Given these values, one can also modify the dendrogram by combining non-significant branches. By adjusting the cut-off level of significance for branches, one can produce dendrograms with a desired level of detail for ease of interpretation. We demonstrate the usefulness of this approach by applying it to illustrative data sets.

Chromatin immunoprecipitation followed by sequencing (ChIP-seq) is a technique for genome-wide profiling of DNA-binding proteins, histone modifications or nucleosomes. Owing to the tremendous progress in next-generation sequencing technology, ChIP-seq offers higher resolution, less noise and greater coverage than its array-based predecessor ChIP-chip. With the decreasing cost of sequencing, ChIP-seq has become an indispensable tool for studying gene regulation and epigenetic mechanisms. In this Review, I describe the benefits and challenges in harnessing this technique with an emphasis on issues related to experimental design and data analysis. ChIP-seq experiments generate large quantities of data, and effective computational analysis will be crucial for uncovering biological mechanisms.

Eukaryotic DNA is wrapped around a histone protein core to constitute the fundamental repeating units of chromatin, the nucleosomes. The affinity of the histone core for DNA depends on the nucleotide sequence; however, it is unclear to what extent DNA sequence determines nucleosome positioning in vivo, and if the same rules of sequence-directed positioning apply to genomes of varying complexity. Using the data generated by high-throughput DNA sequencing combined with chromatin immunoprecipitation, we have identified positions of nucleosomes containing the H2A.Z histone variant and histone H3 trimethylated at lysine 4 in human CD4(+) T-cells. We find that the 10-bp periodicity observed in nucleosomal sequences in yeast and other organisms is not pronounced in human nucleosomal sequences. This result was confirmed for a broader set of mononucleosomal fragments that were not selected for any specific histone variant or modification. We also find that human H2A.Z nucleosomes protect only approximately 120 bp of DNA from MNase digestion and exhibit specific sequence preferences, suggesting a novel mechanism of nucleosome organization for the H2A.Z variant.

MOTIVATION: Type 2 diabetes is a chronic metabolic disease that involves both environmental and genetic factors. To understand the genetics of type 2 diabetes and insulin resistance, the DIabetes Genome Anatomy Project (DGAP) was launched to profile gene expression in a variety of related animal models and human subjects. We asked whether these heterogeneous models can be integrated to provide consistent and robust biological insights into the biology of insulin resistance. RESULTS: We perform integrative analysis of the 16 DGAP data sets that span multiple tissues, conditions, array types, laboratories, species, genetic backgrounds and study designs. For each data set, we identify differentially expressed genes compared with control. Then, for the combined data, we rank genes according to the frequency with which they were found to be statistically significant across data sets. This analysis reveals RetSat as a widely shared component of mechanisms involved in insulin resistance and sensitivity and adds to the growing importance of the retinol pathway in diabetes, adipogenesis and insulin resistance. Top candidates obtained from our analysis have been confirmed in recent laboratory studies.

UNLABELLED: Accurate estimation of DNA copy numbers from array comparative genomic hybridization (CGH) data is important for characterizing the cancer genome. An important part of this process is the segmentation of the log-ratios between the sample and control DNA along the chromosome into regions of different copy numbers. However, multiple algorithms are available in the literature for this procedure and the results can vary substantially among these. Thus, a visualization tool that can display the segmented profiles from a number of methods can be helpful to the biologist or the clinician to ascertain that a feature of interest did not arise as an artifact of the algorithm. Such a tool also allows the methodologist to easily contrast his method against others. We developed a web-based tool that applies a number of popular algorithms to a single array CGH profile entered by the user. It generates a heatmap panel of the segmented profiles for each method as well as a consensus profile. The clickable heatmap can be moved along the chromosome and zoomed in or out. It also displays the time that each algorithm took and provides numerical values of the segmented profiles for download. The web interface calls algorithms written in the statistical language R. We encourage developers of new algorithms to submit their routines to be incorporated into the website. AVAILABILITY: http://compbio.med.harvard.edu/CGHweb.

Recent progress in massively parallel sequencing platforms has enabled genome-wide characterization of DNA-associated proteins using the combination of chromatin immunoprecipitation and sequencing (ChIP-seq). Although a variety of methods exist for analysis of the established alternative ChIP microarray (ChIP-chip), few approaches have been described for processing ChIP-seq data. To fill this gap, we propose an analysis pipeline specifically designed to detect protein-binding positions with high accuracy. Using previously reported data sets for three transcription factors, we illustrate methods for improving tag alignment and correcting for background signals. We compare the sensitivity and spatial precision of three peak detection algorithms with published methods, demonstrating gains in spatial precision when an asymmetric distribution of tags on positive and negative strands is considered. We also analyze the relationship between the depth of sequencing and characteristics of the detected binding positions, and provide a method for estimating the sequencing depth necessary for a desired coverage of protein binding sites.

Throughout development, cell fate decisions are converted into epigenetic information that determines cellular identity. Covalent histone modifications are heritable epigenetic marks and are hypothesized to play a central role in this process. In this report, we assess the concordance of histone H3 lysine 4 dimethylation (H3K4me2) and trimethylation (H3K4me3) on a genome-wide scale in erythroid development by analyzing pluripotent, multipotent, and unipotent cell types. Although H3K4me2 and H3K4me3 are concordant at most genes, multipotential hematopoietic cells have a subset of genes that are differentially methylated (H3K4me2+/me3-). These genes are transcriptionally silent, highly enriched in lineage-specific hematopoietic genes, and uniquely susceptible to differentiation-induced H3K4 demethylation. Self-renewing embryonic stem cells, which restrict H3K4 methylation to genes that contain CpG islands (CGIs), lack H3K4me2+/me3- genes. These data reveal distinct epigenetic regulation of CGI and non-CGI genes during development and indicate an interactive relationship between DNA sequence and differential H3K4 methylation in lineage-specific differentiation.

Next-generation sequencing is poised to unleash dramatic changes in every area of molecular biology. In the past few years, chromatin immunoprecipitation (ChIP) on tiled microarrays (ChIP-chip) has been an important tool for genome-wide mapping of DNA-binding proteins or histone modifications. Now, ChIP followed by direct sequencing of DNA fragments (ChIP-seq) offers superior data with less noise and higher resolution and is likely to replace ChIP-chip in the near future. We will describe advantages of this new technology and outline some of the issues in dealing with the data. ChIP-seq generates considerably larger quantities of data and the most challenging aspect for investigators will be computational and statistical analysis necessary to uncover biological insights hidden in the data.

Array comparative genomic hybridization (aCGH) is a technique for measuring chromosomal aberrations in genomic DNA. With the availability of high-resolution microarrays, detailed characterization of the cancer genome has become possible. In this review, we discuss several issues in the generation and interpretation of aCGH data, including array platforms, experimental design, and data analysis. Due to the complexity of the data, application of appropriate statistical methods is crucial for avoiding false positive findings. We also describe integration of copy number data with other types of data to identify functional significance of observed aberrations.

MOTIVATION: DNA copy number aberrations (CNAs) and gene expression (GE) changes provide valuable information for studying chromosomal instability and its consequences in cancer. While it is clear that the structural aberrations and the transcript levels are intertwined, their relationship is more complex and subtle than initially suspected. Most studies so far have focused on how a CNA affects the expression levels of those genes contained within that CNA. RESULTS: To better understand the impact of CNAs on expression, we investigated the correlation of each CNA to all other genes in the genome. The correlations are computed over multiple patients that have both expression and copy number measurements in brain, bladder and breast cancer data sets. We find that a CNA has a direct impact on the gene amplified or deleted, but it also has a broad, indirect impact elsewhere. To identify a set of CNAs that is coordinately associated with the expression changes of a set of genes, we used a biclustering algorithm on the correlation matrix. For each of the three cancer types examined, the aberrations in several loci are associated with cancer-type specific biological pathways that have been described in the literature: CNAs of chromosome (chr) 7p13 were significantly correlated with epidermal growth factor receptor signaling pathway in glioblastoma multiforme, chr 13q with NF-kappaB cascades in bladder cancer, and chr 11p with Reck pathway in breast cancer. In all three data sets, gene sets related to cell cycle/division such as M phase, DNA replication and cell division were also associated with CNAs. Our results suggest that CNAs are both directly and indirectly correlated with changes in expression and that it is beneficial to examine the indirect effects of CNAs. AVAILABILITY: The code is available upon request.

The distribution of nucleosomes along the genome is a significant aspect of chromatin structure and is thought to influence gene regulation through modulation of DNA accessibility. However, properties of nucleosome organization remain poorly understood, particularly in mammalian genomes. Toward this goal we used tiled microarrays to identify stable nucleosome positions along the HOX gene clusters in human cell lines. We show that nucleosome positions exhibit sequence properties and long-range organization that are different from those characterized in other organisms. Despite overall variability of internucleosome distances, specific loci contain regular nucleosomal arrays with 195-bp periodicity. Moreover, such arrays tend to occur preferentially toward the 3' ends of genes. Through comparison of different cell lines, we find that active transcription is correlated with increased positioning of nucleosomes, suggesting an unexpected role for transcription in the establishment of well-positioned nucleosomes.

SUMMARY: Sequence-directed mapping of nucleosome positions is of major biological interest. Here, we present a web-interface for estimation of the affinity of the histone core to DNA and prediction of nucleosome arrangement on a given sequence. Our approach is based on assessment of the energy cost of imposing the deformations required to wrap DNA around the histone surface. The interface allows the user to specify a number of options such as selecting from several structural templates for threading calculations and adding random sequences to the analysis. AVAILABILITY: The nuScore interface is freely available for use at http://compbio.med.harvard.edu/nuScore. CONTACT: peter_park@harvard.edu; tolstorukov@gmail.com SUPPLEMENTARY INFORMATION: The site contains user manual, description of the methodology and examples.

The Drosophila MSL complex associates with active genes specifically on the male X chromosome to acetylate histone H4 at lysine 16 and increase expression approximately 2-fold. To date, no DNA sequence has been discovered to explain the specificity of MSL binding. We hypothesized that sequence-specific targeting occurs at "chromatin entry sites," but the majority of sites are sequence independent. Here we characterize 150 potential entry sites by ChIP-chip and ChIP-seq and discover a GA-rich MSL recognition element (MRE). The motif is only slightly enriched on the X chromosome ( approximately 2-fold), but this is doubled when considering its preferential location within or 3' to active genes (>4-fold enrichment). When inserted on an autosome, a newly identified site can direct local MSL spreading to flanking active genes. These results provide strong evidence for both sequence-dependent and -independent steps in MSL targeting of dosage compensation to the male X chromosome.

BACKGROUND: Chromatin immunoprecipitation on tiling arrays (ChIP-chip) has been widely used to investigate the DNA binding sites for a variety of proteins on a genome-wide scale. However, several issues in the processing and analysis of ChIP-chip data have not been resolved fully, including the effect of background (mock control) subtraction and normalization within and across arrays. RESULTS: The binding profiles of Drosophila male-specific lethal (MSL) complex on a tiling array provide a unique opportunity for investigating these topics, as it is known to bind on the X chromosome but not on the autosomes. These large bound and control regions on the same array allow clear evaluation of analytical methods.We introduce a novel normalization scheme specifically designed for ChIP-chip data from dual-channel arrays and demonstrate that this step is critical for correcting systematic dye-bias that may exist in the data. Subtraction of the mock (non-specific antibody or no antibody) control data is generally needed to eliminate the bias, but appropriate normalization obviates the need for mock experiments and increases the correlation among replicates. The idea underlying the normalization can be used subsequently to estimate the background noise level in each array for normalization across arrays. We demonstrate the effectiveness of the methods with the MSL complex binding data and other publicly available data. CONCLUSION: Proper normalization is essential for ChIP-chip experiments. The proposed normalization technique can correct systematic errors and compensate for the lack of mock control data, thus reducing the experimental cost and producing more accurate results.

X-chromosome dosage compensation in Drosophila requires the male-specific lethal (MSL) complex, which up-regulates gene expression from the single male X chromosome. Here, we define X-chromosome-specific MSL binding at high resolution in two male cell lines and in late-stage embryos. We find that the MSL complex is highly enriched over most expressed genes, with binding biased toward the 3' end of transcription units. The binding patterns are largely similar in the distinct cell types, with approximately 600 genes clearly bound in all three cases. Genes identified as clearly bound in one cell type and not in another indicate that attraction of MSL complex correlates with expression state. Thus, sequence alone is not sufficient to explain MSL targeting. We propose that the MSL complex recognizes most X-linked genes, but only in the context of chromatin factors or modifications indicative of active transcription. Distinguishing expressed genes from the bulk of the genome is likely to be an important function common to many chromatin organizing and modifying activities.

MOTIVATION: Several statistical methods that combine analysis of differential gene expression with biological knowledge databases have been proposed for a more rapid interpretation of expression data. However, most such methods are based on a series of univariate statistical tests and do not properly account for the complex structure of gene interactions. RESULTS: We present a simple yet effective multivariate statistical procedure for assessing the correlation between a subspace defined by a group of genes and a binary phenotype. A subspace is deemed significant if the samples corresponding to different phenotypes are well separated in that subspace. The separation is measured using Hotelling's T(2) statistic, which captures the covariance structure of the subspace. When the dimension of the subspace is larger than that of the sample space, we project the original data to a smaller orthonormal subspace. We use this method to search through functional pathway subspaces defined by Reactome, KEGG, BioCarta and Gene Ontology. To demonstrate its performance, we apply this method to the data from two published studies, and visualize the results in the principal component space.

A novel genome-wide screen that combines patient outcome analysis with array comparative genomic hybridization and mRNA expression profiling was developed to identify genes with copy number alterations, aberrant mRNA expression, and relevance to survival in glioblastoma. The method led to the discovery of physical gene clusters within the cancer genome with boundaries defined by physical proximity, correlated mRNA expression patterns, and survival relatedness. These boundaries delineate a novel genomic interval called the functional common region (FCR). Many FCRs contained genes of high biological relevance to cancer and were used to pinpoint functionally significant DNA alterations that were too small or infrequent to be reliably identified using standard algorithms. One such FCR contained the EphA2 receptor tyrosine kinase. Validation experiments showed that EphA2 mRNA overexpression correlated inversely with patient survival in a panel of 21 glioblastomas, and ligand-mediated EphA2 receptor activation increased glioblastoma proliferation and tumor growth via a mitogen-activated protein kinase-dependent pathway. This novel genome-wide approach greatly expanded the list of target genes in glioblastoma and represents a powerful new strategy to identify the upstream determinants of tumor phenotype in a range of human cancers.

Accurate and rapid identification of perturbed pathways through the analysis of genome-wide expression profiles facilitates the generation of biological hypotheses. We propose a statistical framework for determining whether a specified group of genes for a pathway has a coordinated association with a phenotype of interest. Several issues on proper hypothesis-testing procedures are clarified. In particular, it is shown that the differences in the correlation structure of each set of genes can lead to a biased comparison among gene sets unless a normalization procedure is applied. We propose statistical tests for two important but different aspects of association for each group of genes. This approach has more statistical power than currently available methods and can result in the discovery of statistically significant pathways that are not detected by other methods. This method is applied to data sets involving diabetes, inflammatory myopathies, and Alzheimer's disease, using gene sets we compiled from various public databases. In the case of inflammatory myopathies, we have correctly identified the known cytotoxic T lymphocyte-mediated autoimmunity in inclusion body myositis. Furthermore, we predicted the presence of dendritic cells in inclusion body myositis and of an IFN-alpha/beta response in dermatomyositis, neither of which was previously described. These predictions have been subsequently corroborated by immunohistochemistry.

MOTIVATION: Array Comparative Genomic Hybridization (CGH) can reveal chromosomal aberrations in the genomic DNA. These amplifications and deletions at the DNA level are important in the pathogenesis of cancer and other diseases. While a large number of approaches have been proposed for analyzing the large array CGH datasets, the relative merits of these methods in practice are not clear. RESULTS: We compare 11 different algorithms for analyzing array CGH data. These include both segment detection methods and smoothing methods, based on diverse techniques such as mixture models, Hidden Markov Models, maximum likelihood, regression, wavelets and genetic algorithms. We compute the Receiver Operating Characteristic (ROC) curves using simulated data to quantify sensitivity and specificity for various levels of signal-to-noise ratio and different sizes of abnormalities. We also characterize their performance on chromosomal regions of interest in a real dataset obtained from patients with Glioblastoma Multiforme. While comparisons of this type are difficult due to possibly sub-optimal choice of parameters in the methods, they nevertheless reveal general characteristics that are helpful to the biological investigator.