The soundness of clustering in the analysis of gene expression profiles and gene function prediction is based on the hypothesis that genes with similar expression profiles may imply strong correlations with their functions in the biological activities. Gene Ontology (GO) has become a well accepted standard in organizing gene function categories. Different gene function categories in GO can have very sophisticated relationships, such as ’part of ’ and ’overlapping’. Until now, no clustering algorithm can generate gene clusters within which the relationships can naturally reflect those of gene function categories in the GO hierarchy. The failure in resembling the relationships may reduce the confidence of clustering in gene function prediction. In this paper, we present a new clustering

Clustering is a common methodology for analyzing the gene expression data. In this paper, we present a new clustering algorithm from an information-theoretic point of view. First, we propose the minimum entropy (measured on a posteriori probabilities) criterion, which is the conditional entropy of clusters given the observations. Fano’s inequality indicates that it could be a good criterion for clustering. We generalize the criterion by replacing Shannon’s entropy with Havrda-Charvat’s structural α-entropy. Interestingly, the minimum entropy criterion based on structural α-entropy is equal to the probability error of the nearest neighbor method when α =2. This is another evidence that the proposed criterion is good for clustering. With a nonparametric approach for estimating a posteriori probabilities, an efficient iterative algorithm is then established to minimize the entropy. The experimental results show that the clustering algorithm performs significantly better than k-means/medians, hierarchical clustering, SOM, and EM in terms of adjusted Rand index. Particularly, our algorithm performs very well even when the correct number of clusters is unknown. In addition, most clustering algorithms produce poor partitions in presence of outliers while our method can correctly reveal the structure of data and effectively identify outliers simultaneously. 1.

Abstract not found

Natural killer cells constitute 50–90 % of lymphocytes in human uterine decidua in early pregnancy. Here, CD56bright uterine decidual NK (dNK) cells were compared with the CD56bright and CD56dim peripheral NK cell subsets by microarray analysis, with verification of results by flow cytometry and RT-PCR. Among the �10,000 genes studied, 278 genes showed at least a threefold change with P � 0.001 when comparing the dNK and peripheral NK cell subsets, most displaying increased expression in dNK cells. The largest number of these encoded surface proteins, including the unusual lectinlike receptors NKG2E and Ly-49L, several killer cell Iglike receptors, the integrin subunits �D, �X, �1, and �5, and multiple tetraspanins (CD9, CD151, CD53, CD63, and TSPAN-5). Additionally, two secreted proteins, galectin-1 and progestagen-associated protein 14, known to have immunomodulatory functions, were selectively expressed in dNK cells. Key words: pregnancy • maternal-fetal relations • natural killer cells • gene expression profiling • lymphocyte subsets The online version of this article includes supplemental material. Address correspondence to Louise A. Koopman at her present address

Background: Genome projects have provided a vast amount of sequence information. Sequence refereed research deposited research comparison between species helps to establish functional catalogues within organisms and to study how they are maintained and modified across phylogenetic groups during evolution. Microarray studies allow us to determine groups of genes with similar temporal regulation and perhaps also common regulatory upstream regions for binding of transcription factors. The integration of sequence and expression data is expected to refine our current annotations and provide some insight into the evolution of gene regulation across organisms. Results: We have investigated how well the protein subcellular localization and functional categories established from clustering of orthologous genes agree with gene-expression data in Saccharomyces cerevisiae. An increase in the resolution of biologically meaningful classes is observed upon the combination of experiments under different conditions. The functional categories deduced by sequence comparison approaches are, in general, preserved at the level of expression and can sometimes interact into larger co-regulated networks, such as the protein translation process. Differences and similarities in the expression between cytoplasmic-mitochondrial and interspecies translation machineries complement evolutionary information from sequence similarity. Conclusions: Combination of several microarray experiments is a powerful tool for the identification of upstream regulatory motifs of yeast genes involved in protein synthesis. Comparison of these yeast co-regulated genes against the archaeal and bacterial operons indicates that the components of the protein translation process are conserved across organisms at the expression level with minor specific adaptations.

Signal transducer and activator of transcription (STAT) proteins are latent transcription factors that mediate a wide range of actions induced by cytokines, interferons, and growth factors. We now report the development of thymic T cell lymphoblastic lymphomas in transgenic mice in which Stat5a or Stat5b is overexpressed within the lymphoid compartment. The rate of lymphoma induction was markedly enhanced by immunization or by the introduction of TCR transgenes. Remarkably, the Stat5 transgene potently induced development of CD8 � T cells, even in mice expressing a class II–restricted TCR transgene, with resulting CD8 � T cell lymphomas. These data demonstrate the oncogenic potential of dysregulated expression of a STAT protein that is not constitutively activated, and that TCR stimulation can contribute to this process. Key words:

Operations research, the science of decision-making, offers a logical structure to management decisions. Operations research is known by several synonyms including systems analysis, quantitative analysis, and management science. Systems decisions can be classified as allocation, scheduling and routing, competition, queuing (waiting) lines, inventory control, searching, replacement, and maintenance. Operations research offers tools such as linear programming, nonlinear programming, decision analysis, neural networks, genetic algorithms, etc., to approach those decisions in an efficient and effective manner. An example from operations research literature is given for the successful application of systems analysis to optimize products and profits at Sadia, the largest poultry producer in Brazil.

B cell–derived chronic lymphocytic leukemia (B-CLL) represents a common malignancy whose cell derivation and pathogenesis are unknown. Recent studies have shown that �50% of CLLs display hypermutated immunoglobulin variable region (IgV) sequences and a more favorable prognosis, suggesting that they may represent a distinct subset of CLLs which have transited through germinal centers (GCs), the physiologic site of IgV hypermutation. To further investigate the phenotype of CLLs, their cellular derivation and their relationship to normal B cells, we have analyzed their gene expression profiles using oligonucleotide-based DNA chip microarrays representative of �12,000 genes. The results show that CLLs display a common and characteristic gene expression profile that is largely independent of their IgV genotype. Nevertheless, a restricted number of genes (�30) have been identified whose differential expression can distinguish IgV mutated versus unmutated cases and identify them in independent panels of cases. Comparison of CLL profiles with those of purified normal B cell subpopulations indicates that the common CLL profile is more related to memory B cells than to those derived from naive B cells, CD5 � B cells, and GC centroblasts and centrocytes. Finally, this analysis has identified a subset of genes specifically expressed by CLL cells of potential pathogenetic and clinical relevance. Key words: somatic hypermutation • germinal center • CD5 • DNA microarray • cluster analysis

Hairy cell leukemia (HCL) is a chronic B cell malignancy characterized by the diffuse infiltration of bone marrow and spleen by cells displaying a typical “hairy ” morphology. However, the nature of the HCL phenotype and its relationship to normal B cells and to other lymphoma subtypes remains unclear. Using gene expression profiling, we show here that HCL displays a homogeneous pattern of gene expression, which is clearly distinct from that of other B cell non-Hodgkin lymphomas. Comparison with the gene expression profiles of purified normal B cell subpopulations, including germinal center (GC), pre-GC (naive), and post-GC (memory) B cells, shows that HCL cells are more related to memory cells, suggesting a derivation from this B cell population. Notably, when compared with memory cells, HCL cells displayed a remarkable conservation in proliferation, apoptosis, and DNA metabolism programs, whereas they appeared significantly altered in the expression of genes controlling cell adhesion and response to chemokines. Finally, these analyses have identified several genes that are specifically expressed in HCL and whose expression was confirmed at the protein level by immunocytochemical analysis of primary HCL cases. These results have biological implications relevant to the pathogenesis of this malignancy as well as clinical implications for its diagnosis and therapy. Key words: DNA microarray • germinal center • hairy morphology • marrow fibrosis • homing

11Department of Oncology, and