Class discovery and classification of tumor samples using mixture modeling of gene expression data--a unified approach

MOTIVATION: The DNA microarray technology has been increasingly used in cancer research. In the literature, discovery of putative classes and classification to known classes based on gene expression data have been largely treated as separate problems. This paper offers a unified approach to class discovery and classification, which we believe is more appropriate, and has greater applicability, in practical situations. RESULTS: We model the gene expression profile of a tumor sample as from a finite mixture distribution, with each component characterizing the gene expression levels in a class. The proposed method was applied to a leukemia dataset, and good results are obtained. With appropriate choices of genes and preprocessing method, the number of leukemia types and subtypes is correctly inferred, and all the tumor samples are correctly classified into their respective type/subtype. Further evaluation of the method was carried out on other variants of the leukemia data and a colon dataset.     

Cluster-Rasch models for microarray gene expression data

Background

We propose two different formulations of the Rasch statistical models to the problem of relating gene expression profiles to the phenotypes. One formulation allows us to investigate whether a cluster of genes with similar expression profiles is related to the observed phenotypes; this model can also be used for future prediction. The other formulation provides an alternative way of identifying genes that are over- or underexpressed from their expression levels in tissue or cell samples of a given tissue or cell type.

Results

We illustrate the methods on available datasets of a classification of acute leukemias and of 60 cancer cell lines. For tumor classification, the results are comparable to those previously obtained. For the cancer cell lines dataset, we found four clusters of genes that are related to drug response for many of the 90 drugs that we considered. In addition, for each type of cell line, we identified genes that are over- or underexpressed relative to other genes.

Conclusions

The cluster-Rasch model provides a probabilistic model for describing gene expression patterns across samples and can be used to relate gene expression profiles to phenotypes.     

Meta-analysis of differentiating mouse embryonic stem cell gene expression kinetics reveals early change of a small gene set

Stem cell differentiation involves critical changes in gene expression. Identification of these should provide endpoints useful for optimizing stem cell propagation as well as potential clues about mechanisms governing stem cell maintenance. Here we describe the results of a new meta-analysis methodology applied to multiple gene expression datasets from three mouse embryonic stem cell (ESC) lines obtained at specific time points during the course of their differentiation into various lineages. We developed methods to identify genes with expression changes that correlated with the altered frequency of functionally defined, undifferentiated ESC in culture. In each dataset, we computed a novel statistical confidence measure for every gene which captured the certainty that a particular gene exhibited an expression pattern of interest within that dataset. This permitted a joint analysis of the datasets, despite the different experimental designs. Using a ranking scheme that favored genes exhibiting patterns of interest, we focused on the top 88 genes whose expression was consistently changed when ESC were induced to differentiate. Seven of these (103728_at, 8430410A17Rik, Klf2, Nr0b1, Sox2, Tcl1, and Zfp42) showed a rapid decrease in expression concurrent with a decrease in frequency of undifferentiated cells and remained predictive when evaluated in additional maintenance and differentiating protocols. Through a novel meta-analysis, this study identifies a small set of genes whose expression is useful for identifying changes in stem cell frequencies in cultures of mouse ESC. The methods and findings have broader applicability to understanding the regulation of self-renewal of other stem cell types.     

Loss of transgelin in breast and colon tumors and in RIE-1 cells by Ras deregulation of gene expression through Raf-independent pathways.

Activated Ras but not Raf can transform RIE-1 and other epithelial cells, indicating the critical importance of Raf-independent effector function in Ras transformation of epithelial cells. To elucidate the nature of these Raf-independent activities, we utilized representational difference analysis to identify genes aberrantly expressed by Ras through Raf-independent mechanisms in RIE-1 cells. We identified a total of 22 genes, both known and novel, whose expression was either activated or abolished by Ras but not Raf. The genes up-regulated encode proteins involved in protein or DNA synthesis, regulation of protease activity, or ligand binding, whereas those genes down-regulated encode actin cytoskeletal-, extracellular matrix-, and gap junction-associated proteins, and transmembrane receptor- or cytokine-like proteins. These results suggest that a key function of Raf-independent signaling involves deregulation of gene expression. We further characterized transgelin as a gene whose expression was abolished by Ras. Transgelin was identified previously as a protein whose expression was lost in virally transformed cell lines. We show that this loss is regulated at the level of gene expression and that both Raf-dependent and Raf-independent pathways are required to cause Ras down-regulation of transgelin in RIE-1 cells, whereas Raf alone is sufficient to cause its loss in NIH 3T3 fibroblasts. We also found that Ras-dependent and Ras-independent mechanisms can cause the down-regulation of transgelin in human breast and colon carcinoma cells lines and patient-derived tumor samples. We conclude that loss of transgelin gene expression may be an important early event in tumor progression and a diagnostic marker for breast and colon cancer development.     

Systematic comparison of gene expression between murine memory and naive B cells demonstrates that memory B cells have unique signaling capabilities

Memory B cells play essential roles in the maintenance of long-term immunity and may be important in the pathogenesis of autoimmune disease, but how these cells are distinguished from their naive precursors is poorly understood. To address this, it would be important to understand how gene expression differs between memory and naive B cells to elucidate memory-specific functions. Using model systems that help overcome the lack of murine memory-specific markers and the low frequency of Ag-specific memory and naive cells, we undertook a global comparison of gene expression between memory B cells and their naive precursors. We identified genes with differential expression and confirmed the differential expression of many of these by quantitative RT-PCR and of some of these at the protein level. Our initial analysis revealed differential expression patterns of genes that regulate signaling. Memory B cells have increased expression of genes important in regulating adenosine signaling and in modulating cAMP responses. Furthermore, memory B cells up-regulate receptors that are essential for embryonic stem cell self-renewal. We further demonstrate that one of these, leukemia inhibitory factor receptor, can initiate functional signaling in memory B cells whereas it does not in naive B cells. Thus, memory and naive B cells are intrinsically wired to signal differently from one another and express a functional signaling pathway that is known to maintain stem cells in other lineages.     

Genes@Work: an efficient algorithm for pattern discovery and multivariate feature selection in gene expression data

MOTIVATION: Despite the growing literature devoted to finding differentially expressed genes in assays probing different tissues types, little attention has been paid to the combinatorial nature of feature selection inherent to large, high-dimensional gene expression datasets. New flexible data analysis approaches capable of searching relevant subgroups of genes and experiments are needed to understand multivariate associations of gene expression patterns with observed phenotypes. RESULTS: We present in detail a deterministic algorithm to discover patterns of multivariate gene associations in gene expression data. The patterns discovered are differential with respect to a control dataset. The algorithm is exhaustive and efficient, reporting all existent patterns that fit a given input parameter set while avoiding enumeration of the entire pattern space. The value of the pattern discovery approach is demonstrated by finding a set of genes that differentiate between two types of lymphoma. Moreover, these genes are found to behave consistently in an independent dataset produced in a different laboratory using different arrays, thus validating the genes selected using our algorithm. We show that the genes deemed significant in terms of their multivariate statistics will be missed using other methods. AVAILABILITY: Our set of pattern discovery algorithms including a user interface is distributed as a package called Genes@Work. This package is freely available to non-commercial users and can be downloaded from our website (http://www.research.ibm.com/FunGen).     

Differential regulation of CDX1 and CDX2 gene expression by deficiency in methyl group donors

The CDX2 and CDX1 homeobox genes have respectively a tumour suppressor and proliferative role in the intestinal epithelium. We analyzed DNA methylation and histones modifications associated with CDX2 and CDX1 promoters in two human colon cancer cell lines expressing differentially these genes, Caco2/TC7 [CDX2 positive-CDX1 negative] and HT29 [CDX2 negative-CDX1 negative] cells. Chromatin immunoprecipitation experiments indicated that CDX2 and CDX1 gene expression correlated with a histone modifications pattern characterizing active chromatin (H3K4 trimethylated and H3 acetylated). Bisulfite DNA sequencing and methylation-specific PCR showed that CDX2 and CDX1 promoters display no methylation in HT29 cells even though both genes are not expressed. In contrast, the CDX1 promoter is methylated in Caco2/TC7. DNA demethylation by 5aza-dC or the combination of 5aza-dC plus SAHA, an inhibitor of histone deacetylases, restored CDX1 expression in Caco2/TC7 cells but these treatments were inefficient on both CDX2 and CDX1 in HT29 cells. Thus, in colon cancer cells the changes in chromatin conformation are heterogeneous and repression of CDX2 and CDX1 in HT29 cells is not due to epigenetic mechanisms. In vivo, dietary deprivation of methyl groups in rats upregulated CDX1 mRNA and downregulated to a lesser extent CDX2 mRNA expression. Moreover, methyl group deprivation downregulated CDX2 protein by changing its phosphorylation pattern. The changes in CDX2 and CDX1 expression determined by methyl group deprivation may constitute one of the mechanisms sustaining the protective role attributed to folate in colon cancer.     

Variation in gene expression patterns in human gastric cancers

Gastric cancer is the world's second most common cause of cancer death. We analyzed gene expression patterns in 90 primary gastric cancers, 14 metastatic gastric cancers, and 22 nonneoplastic gastric tissues, using cDNA microarrays representing approximately 30,300 genes. Gastric cancers were distinguished from nonneoplastic gastric tissues by characteristic differences in their gene expression patterns. We found a diversity of gene expression patterns in gastric cancer, reflecting variation in intrinsic properties of tumor and normal cells and variation in the cellular composition of these complex tissues. We identified several genes whose expression levels were significantly correlated with patient survival. The variations in gene expression patterns among cancers in different patients suggest differences in pathogenetic pathways and potential therapeutic strategies.     

Comparison of computational models for assessing conservation of gene expression across species

Assessing conservation/divergence of gene expression across species is important for the understanding of gene regulation evolution. Although advances in microarray technology have provided massive high-dimensional gene expression data, the analysis of such data is still challenging. To date, assessing cross-species conservation of gene expression using microarray data has been mainly based on comparison of expression patterns across corresponding tissues, or comparison of co-expression of a gene with a reference set of genes. Because direct and reliable high-throughput experimental data on conservation of gene expression are often unavailable, the assessment of these two computational models is very challenging and has not been reported yet. In this study, we compared one corresponding tissue based method and three co-expression based methods for assessing conservation of gene expression, in terms of their pair-wise agreements, using a frequently used human-mouse tissue expression dataset. We find that 1) the co-expression based methods are only moderately correlated with the corresponding tissue based methods, 2) the reliability of co-expression based methods is affected by the size of the reference ortholog set, and 3) the corresponding tissue based methods may lose some information for assessing conservation of gene expression. We suggest that the use of either of these two computational models to study the evolution of a gene's expression may be subject to great uncertainty, and the investigation of changes in both gene expression patterns over corresponding tissues and co-expression of the gene with other genes is necessary.     

Testing association of a pathway with survival using gene expression data

MOTIVATION: A recent surge of interest in survival as the primary clinical endpoint of microarray studies has called for an extension of the Global Test methodology to survival. RESULTS: We present a score test for association of the expression profile of one or more groups of genes with a (possibly censored) survival time. Groups of genes may be pathways, areas of the genome, clusters from a cluster analysis or all genes on a chip. The test allows one to test hypotheses about the influence of these groups of genes on survival directly, without the intermediary of single gene testing. The test is based on the Cox proportional hazards model and is calculated using martingale residuals. It is possible to adjust the test for the presence of covariates. We also present a diagnostic graph to assist in the interpretation of the test result, visualizing the influence of genes. The test is applied to a tumor dataset, revealing pathways from the gene ontology database that are associated with survival of patients. AVAILABILITY: The Global Test for survival has been incorporated into the R-package globaltest (version 3.0), available at http://www.bioconductor.org     

Inflammation-associated DNA methylation patterns in epithelium of ulcerative colitis

Aberrant DNA methylation patterns have been reported in inflamed tissues and may play a role in disease. We studied DNA methylation and gene expression profiles of purified intestinal epithelial cells from ulcerative colitis patients, comparing inflamed and non-inflamed areas of the colon. We identified 577 differentially methylated sites (false discovery rate <0.2) mapping to 210 genes. From gene expression data from the same epithelial cells, we identified 62 differentially expressed genes with increased expression in the presence of inflammation at prostate cancer susceptibility genes PRAC1 and PRAC2. Four genes showed inverse correlation between methylation and gene expression; ROR1, GXYLT2, FOXA2, and, notably, RARB, a gene previously identified as a tumor suppressor in colorectal adenocarcinoma as well as breast, lung and prostate cancer. We highlight targeted and specific patterns of DNA methylation and gene expression in epithelial cells from inflamed colon, while challenging the importance of epithelial cells in the pathogenesis of chronic inflammation.     

CoXpress: differential co-expression in gene expression data

Background  

Traditional methods of analysing gene expression data often include a statistical test to find differentially expressed genes, or use of a clustering algorithm to find groups of genes that behave similarly across a dataset. However, these methods may miss groups of genes which form differential co-expression patterns under different subsets of experimental conditions. Here we describe coXpress, an R package that allows researchers to identify groups of genes that are differentially co-expressed.     

Determination of stromal signatures in breast carcinoma

Many soft tissue tumors recapitulate features of normal connective tissue. We hypothesize that different types of fibroblastic tumors are representative of different populations of fibroblastic cells or different activation states of these cells. We examined two tumors with fibroblastic features, solitary fibrous tumor (SFT) and desmoid-type fibromatosis (DTF), by DNA microarray analysis and found that they have very different expression profiles, including significant differences in their patterns of expression of extracellular matrix genes and growth factors. Using immunohistochemistry and in situ hybridization on a tissue microarray, we found that genes specific for these two tumors have mutually specific expression in the stroma of nonneoplastic tissues. We defined a set of 786 gene spots whose pattern of expression distinguishes SFT from DTF. In an analysis of DNA microarray gene expression data from 295 previously published breast carcinomas, we found that expression of this gene set defined two groups of breast carcinomas with significant differences in overall survival. One of the groups had a favorable outcome and was defined by the expression of DTF genes. The other group of tumors had a poor prognosis and showed variable expression of genes enriched for SFT type. Our findings suggest that the host stromal response varies significantly among carcinomas and that gene expression patterns characteristic of soft tissue tumors can be used to discover new markers for normal connective tissue cells.