Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “unsupervised clustering”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 343 records · Page 19Linked to original sources
PubMed2004

Similarity of mRNA phenotypes of morphologically normal macular and peripheral retinal pigment epithelial cells in older human eyes.

PURPOSE: To determine the expression profiles of morphologically normal human retinal pigment epithelial (RPE) cells that originate from the macula and periphery. METHODS: Morphologically normal RPE cells from 15 human globes from donors aged 52 to 82 years old were laser capture microdissected. Total RNA from 5000 cells was SMART amplified, [33]P-labeled, and hybridized to a cDNA array containing 4325 known genes. Expression profiles were analyzed by hierarchical cluster analysis, Prediction Analysis of Microarrays (PAM), and Significance Analysis for Microarrays (SAM). Differentially expressed genes were evaluated further by real time RT-PCR. RESULTS: The overall expression profiles of RPE cells from the macula and periphery were similar. Unsupervised and supervised hierarchical cluster analysis showed that patient genotype was a stronger separating factor than topographical location. SAM analysis identified 11 genes that were underexpressed by macular RPE cells. The expression patterns of these 11 genes were confirmed by real time RT-PCR, with 5 genes reaching statistical significance. CONCLUSIONS: Whereas the overall expression profiles were similar between cells from the macula and periphery, subtle differential expression of five genes could contribute to RPE phenotypic differences based on topographic location.

Aged↗
PubMed2004

Cluster analysis of soft X-ray spectromicroscopy data.

Soft X-ray spectromicroscopy provides spectral data on the chemical speciation of light elements at sub-100 nm spatial resolution. When all chemical species in a specimen are known and separately characterized, existing approaches can be used to measure the concentration of each component at each pixel. In other cases (such as often occur in biology or environmental science), some spectral signatures may not be known in advance so other approaches must be used. We describe here an approach that uses principal component analysis to orthogonalize and noise-filter spectromicroscopy data. We then use cluster analysis (a form of unsupervised pattern matching) to classify pixels according to spectral similarity, to extract representative, cluster-averaged spectra with good signal-to-noise ratio, and to obtain gradations of concentration of these representative spectra at each pixel. The method is illustrated with a simulated data set of organic compounds, and a mixture of lutetium in hematite used to understand colloidal transport properties of radionuclides.

Cluster Analysis↗
PubMed2004

A fast sequential image fractal coding approach based on optimal fuzzy clustering.

To reduce the coding time of the conventional method, a fast sequential image fractal compression algorithm was proposed on the basis of the principle of optimal fuzzy clustering (OFC) for an unsupervised sample set with the category number settled by the algorithm itself. We utilized the cost function defined by the OFC algorithm to obtain the best category number corresponding to the minimum value of the function. Firstly the Linde-Buzo-Gray (LBG) algorithm was realized to acquire a rough cluster of the domain pool. Then the optimal category number was obtained by implementing our algorithm with small computational cost. Finally the more precise category was gained and the detail of the reconstructed image efficiently preserved. As a global optimal algorithm, OFC not only helps LBG eliminate the local minima, but also effectively compensates for the arbitrary interference in hard clustering problem. Soft clustering of the domain blocks allows classified searches instead of global ones and takes less coding time, and therefore clearly outperforms to the classic method relying on reduction of the size of the domain pool by classification. In computer simulation, OFC-based algorithm for the fractal coding scheme achieved excellent performance. For some standard and sequential medical images, the results denoted that the encoding speed was improved by about 5 folds without affecting the signal-to-noise ratio and compression ratio, and the quality of the reconstructed image could be better retained.

Algorithms↗
PubMed2006

An unsupervised classification scheme for improving predictions of prokaryotic TIS.

BACKGROUND: Although it is not difficult for state-of-the-art gene finders to identify coding regions in prokaryotic genomes, exact prediction of the corresponding translation initiation sites (TIS) is still a challenging problem. Recently a number of post-processing tools have been proposed for improving the annotation of prokaryotic TIS. However, inherent difficulties of these approaches arise from the considerable variation of TIS characteristics across different species. Therefore prior assumptions about the properties of prokaryotic gene starts may cause suboptimal predictions for newly sequenced genomes with TIS signals differing from those of well-investigated genomes. RESULTS: We introduce a clustering algorithm for completely unsupervised scoring of potential TIS, based on positionally smoothed probability matrices. The algorithm requires an initial gene prediction and the genomic sequence of the organism to perform the reannotation. As compared with other methods for improving predictions of gene starts in bacterial genomes, our approach is not based on any specific assumptions about prokaryotic TIS. Despite the generality of the underlying algorithm, the prediction rate of our method is competitive on experimentally verified test data from E. coli and B. subtilis. Regarding genomes with high G+C content, in contrast to some previously proposed methods, our algorithm also provides good performance on P. aeruginosa, B. pseudomallei and R. solanacearum. CONCLUSION: On reliable test data we showed that our method provides good results in post-processing the predictions of the widely-used program GLIMMER. The underlying clustering algorithm is robust with respect to variations in the initial TIS annotation and does not require specific assumptions about prokaryotic gene starts. These features are particularly useful on genomes with high G+C content. The algorithm has been implemented in the tool "TICO" (TIs COrrector) which is publicly available from our web site.

Algorithms↗
PubMed2004

Management of osteoarthritis (OA) with an unsupervised home based exercise programme and/or patient administered assessment tools. A cluster randomised controlled trial with a 2x2 factorial design.

BACKGROUND: Diary recording of pain and disabling activities in osteoarthritis (OA) is widely recommended, but, to our knowledge, its impact on symptoms has not been investigated. Exercise programmes have been shown to be effective when patients are closely supervised by nurses or physiotherapists; however, data are lacking on the efficacy of an unsupervised home based exercise regimen in patients with OA. OBJECTIVES: To evaluate the clinical efficacy of patient administered assessment tools and an unsupervised home based exercise programme alone or in combination in patients with OA. METHODS: The study was a 24 week, open cluster randomised controlled trial with a factorial design. Rheumatologists (n = 867) were assigned to four groups according to the treatment given: standardised tools (ST; n = 220), exercises (EX; n = 213), both tools and exercises (ST+EX; n = 213), or usual care (n = 221). Each rheumatologist was to enroll four patients who met the American College of Rheumatology criteria for OA (three with knee OA, one with hip OA). "Tools" consisted of weekly recording of pain and disabling activities in a diary. A home based exercise programme was performed daily at least four times per week with the aid of videotape and booklet. In addition to exercise and assessment, all patients received 12.5 mg or 25 mg of the non-steroidal anti-inflammatory drug rofecoxib once daily. Outcome variables were: pain (measured on a visual analogue scale, 0-100); Western Ontario and McMaster Universities Osteoarthritis Index, function subscale (0-100); and patient assessment of the quality of care (0-100). RESULTS: Overall, 2957 patients with OA (2216 knee, 741 hip) were included. After 24 weeks, both pain and function improved in the ST, EX, ST+EX, and usual care groups (mean (SD) -17 (27), -20 (29), -15 (27), -19 (29); and -11 (19), -12 (19), -10 (19), -11 (20), respectively), without significant differences between groups. However, patients in the EX and ST+EX groups were more likely to agree that their rheumatologist had done his best to preserve their functional and physical activities. CONCLUSION: Although patients' assessments favoured the exercise programme, results from this study failed to demonstrate a short term symptomatic effect of the two non-pharmacological treatments (weekly recording of condition and exercise) in patients with OA concurrently receiving nonsteroidal anti-inflammatory drugs.

Aged↗
PubMed2005

Clonal relationship of tumor nodules in hepatocellular carcinoma: a hierarchical clustering analysis of comparative genomic hybridization data.

Information on the clonal relationship between tumor nodules in patients diagnosed with hepatocellular carcinoma (HCC) holds prognostic significance in the prediction of recurrence and postoperative treatment. Here, we investigated the clonal relationships in 11 focal nodules from 5 patients with HCC by comparative genomic hybridization (CGH) and spectral karyotyping. CGH analysis indicated tumor nodules in 5 (100%) of 5 cases to share similar patterns of genomic imbalances, suggesting nodules to be clonally related. This clonal relationship was further substantiated in 2 cases, where spectral karyotyping analysis indicated identical structural rearrangements between focal tumors. Few studies have attempted to differentiate multicentric growth from intrahepatic dissemination in HCC using CGH analysis; however, data interpretation remained subjective. In this study, we have extended our data to include published CGH findings on 57 nodules derived from 19 HCC cases. Simulation of CGH findings by unsupervised 2-way hierarchical clustering indicated tumor nodules in 20 (83.3%) of 24 cases to cluster concordantly, signifying a high incidence of clonal similarity. In sum, bioinformatic analysis of genomic profiles may represent a reliable research tool in analyzing clonal relationships among tumor nodules.

Adult↗
PubMed2003

Classification and identification of Arabidopsis cell wall mutants using Fourier-Transform InfraRed (FT-IR) microspectroscopy.

We have developed a novel procedure for the rapid classification and identification of Arabidopsis mutants with altered cell wall architecture based on Fourier-Transform Infrared (FT-IR) microspectroscopy. FT-IR transmission spectra were sampled from native 4-day-old dark-grown hypocotyls of 46 mutants and the wild type treated with various drugs. The Mahalanobis distance between mutants, calculated from the spectral information after compression with the Discriminant Variables Selection procedure, was used for alpha hierarchical cluster analysis. Despite the completely unsupervised nature of the classification procedure, we show that all mutants with cellulose defects appeared in the same cluster. In addition, mutant alleles of similar strength for several unrelated loci were also clustered, which demonstrates the sensitivity of the method to detect a wide array of cell wall defects. Comparing the cellulose-deficient cluster with the cluster that contained wild-type controls led to the identification of wave numbers that were diagnostic for altered cellulose content in the context of an intact cell wall. The results show that FT-IR spectra can be used to identify different classes of mutants and to characterize cell wall changes at a microscopic level in unknown mutants. This procedure significantly accelerates the identification and classification of cell wall mutants, which makes cell wall polysaccharides more accessible to functional genomics approaches.

Alleles↗
PubMed1999

Internal representation in neural networks used for classification of patient anaesthetic states and dosage.

In this study we aimed to explore the ability of artificial neural networks (ANN) to classify patient anaesthetic states and dosage. Surgical data obtained under different states of anaesthesia and dose levels were modelled via this approach. It is shown that inferential parameters can be used to determine the patient anaesthetic states and drug dosage. In addition to demonstrating the capability of ANN for classification we were interested in the internal representations that are developed automatically by networks while they are learning their processing task. An unsupervised learning procedure of clustering via which the classes are inferred from the data and a supervised learning technique of discrimination via which to construct a classification of the known categories were applied to analyse the performance of the ANN. Discriminant analysis (DA) was also utilised to optimise the network architecture.

Anesthesia↗
PubMed2003

Reliable classification of two-class cancer data using evolutionary algorithms.

In the area of bioinformatics, the identification of gene subsets responsible for classifying available disease samples to two or more of its variants is an important task. Such problems have been solved in the past by means of unsupervised learning methods (hierarchical clustering, self-organizing maps, k-mean clustering, etc.) and supervised learning methods (weighted voting approach, k-nearest neighbor method, support vector machine method, etc.). Such problems can also be posed as optimization problems of minimizing gene subset size to achieve reliable and accurate classification. The main difficulties in solving the resulting optimization problem are the availability of only a few samples compared to the number of genes in the samples and the exorbitantly large search space of solutions. Although there exist a few applications of evolutionary algorithms (EAs) for this task, here we treat the problem as a multiobjective optimization problem of minimizing the gene subset size and minimizing the number of misclassified samples. Moreover, for a more reliable classification, we consider multiple training sets in evaluating a classifier. Contrary to the past studies, the use of a multiobjective EA (NSGA-II) has enabled us to discover a smaller gene subset size (such as four or five) to correctly classify 100% or near 100% samples for three cancer samples (Leukemia, Lymphoma, and Colon). We have also extended the NSGA-II to obtain multiple non-dominated solutions discovering as much as 352 different three-gene combinations providing a 100% correct classification to the Leukemia data. In order to have further confidence in the identification task, we have also introduced a prediction strength threshold for determining a sample's belonging to one class or the other. All simulation results show consistent gene subset identifications on three disease samples and exhibit the flexibilities and efficacies in using a multiobjective EA for the gene subset identification task.

Algorithms↗
PubMed2002

Gene expression data analysis of human lymphoma using support vector machines and output coding ensembles.

The large amount of data generated by DNA microarrays was originally analysed using unsupervised methods, such as clustering or self-organizing maps. Recently supervised methods such as decision trees, dot-product support vector machines (SVM) and multi-layer perceptrons (MLP) have been applied in order to classify normal and tumoural tissues. We propose methods based on non-linear SVM with polynomial and Gaussian kernels, and output coding (OC) ensembles of learning machines to separate normal from malignant tissues, to classify different types of lymphoma and to analyse the role of sets of coordinately expressed genes in carcinogenic processes of lymphoid tissues. Using gene expression data from "Lymphochip", a specialised DNA microarray developed at Stanford University School of Medicine, we show that SVM can correctly separate normal from tumoural tissues, and OC ensembles can be successfully used to classify different types of lymphoma. Moreover, we identify a group of coordinately expressed genes related to the separation of two distinct subgroups inside diffuse large B-cell lymphoma (DLBCL), validating a previous Alizadeh's hypothesis about the existence of two distinct diseases inside DLBCL.

Artificial Intelligence↗
PubMed2006

Gene expression profiling of acute promyelocytic leukaemia identifies two subtypes mainly associated with flt3 mutational status.

Acute promyelocytic leukaemia (APL) is a well-defined disease characterized by a typical morphology of leukaemic cells, the presence of t(15;17) translocation and the unique sensitivity to the differentiating effect of all-trans retinoic acid. Nevertheless, some aspects are variable among APL patients, with differences substantially related to morphological variants, peripheral leukocytes count, the presence of a disseminated intravascular coagulopathy, different PML/RARalpha isoforms (long, variable or short) and Fms-like tyrosine kinase 3 (Flt3) mutations. In order to better define this variability, we investigated the gene expression profiles of 18 APL cases revealing, besides a high uniformity in gene expression pattern, the presence of few robust differences among patients able to identify, by an unsupervised analysis, two major clusters of patients characterized by different phenotypes (hypogranular M3v vs classical M3) and by the presence or absence of Flt3 internal tandem duplications (ITDs). Further supervised analysis confirmed that Flt3 status was the APL parameter best associated with these two subgroups. We identified, between Flt3 wild-type and Flt3-ITDs subsets, 147 differentially expressed genes that were involved in the cytoskeleton organization, in the cell adhesion and migration, in the proliferation and the coagulation/inflammation pathways as well as in differentiation and myeloid granules constitution suggesting a role of Flt3 mutations in the pathogenesis and clinical manifestations of APL.

Adult↗
PubMed1996

Kohonen map as a visualization tool for the analysis of protein sequences: multiple alignments, domains and segments of secondary structures.

The method of Kohonen maps, a special form of neural networks, was applied as a visualization tool for the analysis of protein sequence similarity. The procedure converts sequence (domains, aligned sequences, segments of secondary structure) into a characteristic signal matrix. This conversion depends on the property or replacement score vector selected by the user. Similar sequences have small distance in the signal space. The trained Kohonen network is functionally equivalent to an unsupervised non-linear cluster analyzer. Protein families, or aligned sequences, or segments of similar secondary structure, aggregate as clusters, and their proximity may be inspected on a color screen or on paper. Pull-down menus permit access to background information in the established text-oriented way.

Amino Acid Sequence↗
PubMed2005

TICO: a tool for improving predictions of prokaryotic translation initiation sites.

UNLABELLED: We provide the tool 'TICO' (Translation Initiation site COrrection) for improving the results of conventional gene finders for prokaryotic genomes with regard to exact localization of the translation initiation site (TIS). At the current state TICO provides an interface for direct post processing of the predictions obtained from the widely used program GLIMMER. Our program is based on a clustering algorithm for completely unsupervised scoring of potential TIS locations. AVAILABILITY: Our tool can be freely accessed through a web interface at http://tico.gobics.de/ CONTACT: maike@gobics.de

Algorithms↗
PubMed2003

Molecular classification of breast carcinomas using tissue microarrays.

The histopathologic classification of breast cancer stratifies tumors based on tumor grade, stage, and type. Despite an overall correlation with survival, this classification is poorly predictive and tumors with identical grade and stage can have markedly contrasting outcomes. Recently, breast carcinomas have been classified by their gene expression profiles on frozen material. The validation of such a classification on formalin-fixed paraffin-embedded tumor archives linked to clinical information in a high-throughput fashion would have a major impact on clinical practice. The authors tested the ability of tumor tissue microarrays (TMAs) to sub-classify breast cancers using a TMA containing 107 breast cancers. The pattern of expression of 13 different protein biomarkers was assessed by immunohistochemistry and the multidimensional data was analyzed using an unsupervised two-dimensional clustering algorithm. This revealed distinct tumor clusters which divided into two main groups correlating with tumor grade (P<0.001) and nodal status (P = 0.04). None of the protein biomarkers tested could individually identify these groups. The biological significance of this classification is supported by its similarity with one derived from gene expression microarray analysis. Thus, molecular profiling of breast cancer using a limited number of protein biomarkers in TMAs can sub-classify tumors into clinically and biologically relevant subgroups.

Adenocarcinoma↗
PubMed2006

Pilot study of mucosal genetic differences in early smokers and nonsmokers.

BACKGROUND: Global gene expression analysis is proving to be an important means of assessing human tumors and may identify key components of carcinogenesis or clinical prognosis. This technique has been successfully applied to head and neck squamous cell carcinoma (HNSCC) and thyroid carcinomas; however, little has been done to evaluate premalignant states. METHODS: Human buccal mucosal cells were sampled from smokers and nonsmokers using a noninvasive brush technique. The method was validated by assessing the quantity and quality of RNA obtained. The purified RNA was then assayed using cDNA microarrays containing 27,323 cDNA clones to examine the buccal mucosa in these patients for differences in gene expression patterns. Using unsupervised and supervised hierarchical clustering methods, we developed a gene profile signature for an initial training set of smokers and nonsmokers and then used this to predict smoking status in a subsequent test set of subjects. Selected genes were then cross-referenced with previously published gene sets found in HNSCC identified by our group. RESULTS: Nineteen subjects were used in this pilot analysis, 9 smokers and 10 nonsmokers. Smoking among the study group ranged from 1 to 60 pack years. RNA purified from buccal mucosal brushing demonstrated a high degree of similarity in gene expression profiles among independent samples. Through the application of supervised clustering techniques, we were able to identify 113 genes whose expression differed significantly between samples from smokers and nonsmokers (t test, P < .001). This expression signature was able to accurately predict who within the second set of subjects were smokers, with the exception of one person who had a minimal tobacco history and clustered with the nonsmokers. Cross-referencing data with that found in HNSCC, we were able to identify a tumor suppressor gene involved in the c-myc pathway (Mxi1) that was similarly under-expressed in smokers and cancer patients with progressive disease. CONCLUSIONS: Although the sample size was small in this preliminary dataset, our analysis revealed several groups of genes that were either over- or under-expressed in the smokers and which could be used to predict smoking exposure. Many of these represent genes of possible interest as early molecular markers for head and neck carcinogenesis.

Adult↗
PubMed2006

ProtoBee: hierarchical classification and annotation of the honey bee proteome.

The recently sequenced genome of the honey bee (Apis mellifera) has produced 10,157 predicted protein sequences, calling for a computational effort to extract biological insights from them. We have applied an unsupervised hierarchical protein-clustering method, which was previously used in the ProtoNet system, to nearly 200,000 proteins consisting of the predicted honey bee proteins, the SWISS-PROT protein database, and the complete set of proteins of the mouse (Mus musculus) and the fruit fly (Drosophila melanogaster). The hierarchy produced by this method has been entitled ProtoBee. In ProtoBee, the proteins are hierarchically organized into 18,936 separate tree hierarchies, each representing a protein functional family. By using the mouse and Drosophila complete proteomes as reference, we are able to highlight functional groups of putative gene-loss events, putative novel proteins of unique functionality, and bee-specific paralogs. We have studied some of the ProtoBee findings and suggest their biological relevance. Examples include novel opsin genes and intriguing nuclear matches of mitochondrial genes. The organization of bee sequences into functional clusters suggests a natural way of automatically inferring functional annotation. Following this notion, we were able to assign functional annotation to about 70% of the sequences. ProtoBee is available at http://www.protobee.cs.huji.ac.il.

Animals↗
PubMed2003

Gene expression dynamics during germinal center transit in B cells.

The germinal center (GC) reaction in T cell dependent antibody responses is crucial for the generation of B cell memory and plays a critical role in B cell lymphomagenesis. To gain insight into the physiology of this reaction, we identified the transcriptional changes that occur in B cells during the GC-transit (naïve B cells --> CD77(+) centroblasts (CBs) --> CD77(-) centrocytes (CCs) --> memory B cells) by DNA microarray experiments and the subsequent data analysis employing unsupervised and supervised hierarchical clustering. The naïve B cell is characterized by a nonproliferative, anti-apoptotic phenotype and the expression of various chemokine and cytokine receptors. The transition from naïve B cells to CBs is associated with (1) the up-regulation of genes associated with cellular proliferation, DNA-repair, and chromatin remodeling; (2) the acquisition of a pro-apoptotic phenotype; (3) the down-regulation of cytokine, chemokine, and adhesion receptors expressed in the naïve cells; and (4) the expression of a distinct adhesion repertoire. The CB and the CC revealed surprisingly few gene expression differences, suggesting that the CC is heterogeneous in its cellular composition. The CB/CC to memory B cell transition shows a general reversion to the profile characteristic for the naïve B cells, with the exception of the up-regulation of several surface receptors, including CD27, CD80, and IL-2Rbeta, and the simultaneous expression of both anti- and pro-apoptotic genes. These gene expression profiles of the normal B cell subpopulations are being used to identify the signals occurring during GC development, the cellular derivation of various types of B cell malignancies, and the genes deregulated in GC-derived tumors.

B-Lymphocytes↗
PubMed2004

Whole blood and leukocyte RNA isolation for gene expression analyses.

The analysis of gene expression data in clinical medicine has been plagued by the lack of a critical evaluation of accepted methodologies for the collection, processing, and labeling of RNA. In the present report, the reliability of two commonly used techniques to isolate RNA from whole blood or its leukocyte compartment was compared by examining their reproducibility, variance, and signal-to-noise ratios. Whole blood was obtained from healthy subjects and was either untreated or stimulated ex vivo with Staphylococcus enterotoxin B (SEB). Blood samples were also obtained from trauma patients but were not stimulated with SEB ex vivo. Total RNA was isolated from whole blood with the PAXgene proprietary blood collection system or from isolated leukocytes. Biotin-labeled cRNA was hybridized to Affymetrix GeneChips. The Pearson correlation coefficient for gene expression measurements in replicates from healthy subjects with both techniques was excellent, exceeding 0.985. Unsupervised analyses, including hierarchical cluster analysis, however, revealed that the RNA isolation method resulted in greater differences in gene expression than stimulation with SEB or among different trauma patients. The intraclass correlation, a measure of signal-to-noise ratio, of the difference between SEB-stimulated and unstimulated blood from healthy subjects was significantly higher in leukocyte-derived samples than in whole blood: 0.75 vs. 0.46 (P = 0.002). At the P < 0.001 level of significance, twice as many probe sets discriminated between SEB-stimulated and unstimulated blood with leukocyte isolation than with PAXgene. The findings suggest that the method of RNA isolation from whole blood is a critical variable in the design of clinical studies using microarray analyses.

Adolescent↗