SEARCH · PubMed Health

Results for “Clustering Algorithms”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 1,351 records · Page 75Linked to original sources

PubMed1994

Patterns in the complementary determining regions of immunoglobulins (CDRs).

An analysis of the frequency of use of amino acids on the CDR-1 and CDR-2 of 1500 immunoglobulins showed that the frequencies of amino acids in different positions could be fitted by two types of distribution. For some positions the frequencies were fitted by an inverse power law and for other positions by an exponential distribution. In order to see whether the more frequently used amino acids for specific positions had physicochemical properties or attributes in common, they were clustered using an algorithm normally applied to artificial intelligence problems. It was found that the amino acids in those positions fitted by the inverse power law have similar hydrophobicity and volume, which are commonly attributes of amino acids in structural positions. Thus, if these positions are critical to maintaining the structural features of the CDR domains, the rest of the positions should be either properly involved in the recognition process or irrelevant. The frequencies of amino acids in these recognition positions were fitted by the exponential law, and it was found by the clustering analysis that these amino acids share properties of a more general type, such as capability of forming hydrogen bonds, polarity, etc. This suggests that at least part of the recognition mechanism requires general properties rather than specific amino acids. Amino acids sharing the required attributes for each one of these positions are then used with random frequency.

Algorithms↗

PubMed2005

Genomic multiple sequence alignments: refinement using a genetic algorithm.

BACKGROUND: Genomic sequence data cannot be fully appreciated in isolation. Comparative genomics--the practice of comparing genomic sequences from different species--plays an increasingly important role in understanding the genotypic differences between species that result in phenotypic differences as well as in revealing patterns of evolutionary relationships. One of the major challenges in comparative genomics is producing a high-quality alignment between two or more related genomic sequences. In recent years, a number of tools have been developed for aligning large genomic sequences. Most utilize heuristic strategies to identify a series of strong sequence similarities, which are then used as anchors to align the regions between the anchor points. The resulting alignment is globally correct, but in many cases is suboptimal locally. We describe a new program, GenAlignRefine, which improves the overall quality of global multiple alignments by using a genetic algorithm to improve local regions of alignment. Regions of low quality are identified, realigned using the program T-Coffee, and then refined using a genetic algorithm. Because a better COFFEE (Consistency based Objective Function For alignmEnt Evaluation) score generally reflects greater alignment quality, the algorithm searches for an alignment that yields a better COFFEE score. To improve the intrinsic slowness of the genetic algorithm, GenAlignRefine was implemented as a parallel, cluster-based program. RESULTS: We tested the GenAlignRefine algorithm by running it on a Linux cluster to refine sequences from a simulation, as well as refine a multiple alignment of 15 Orthopoxvirus genomic sequences approximately 260,000 nucleotides in length that initially had been aligned by Multi-LAGAN. It took approximately 150 minutes for a 40-processor Linux cluster to optimize some 200 fuzzy (poorly aligned) regions of the orthopoxvirus alignment. Overall sequence identity increased only slightly; but significantly, this occurred at the same time that the overall alignment length decreased--through the removal of gaps--by approximately 200 gapped regions representing roughly 1,300 gaps. CONCLUSION: We have implemented a genetic algorithm in parallel mode to optimize multiple genomic sequence alignments initially generated by various alignment tools. Benchmarking experiments showed that the refinement algorithm improved genomic sequence alignments within a reasonable period of time.

Algorithms↗

PubMed2004

Molecular phylogeny of oribatid mites (Oribatida, Acari): evidence for multiple radiations of parthenogenetic lineages.

Nucleotide sequences of the D3 expansion segment and its flanking regions of the 28S rDNA gene were used to evaluate phylogenetic relationships among representative sexual and asexual oribatid mites (Oribatida, Acariformes). The aim of this study was to investigate the hypothesis that oribatid mites consist of species-rich clusters of asexual species that may have radiated while being parthenogenetic. Furthermore, the systematic position of the astigmate mites (Astigmata, Acariformes) which have been hypothesised to represent a paedomorphic lineage within the oribatid mites, is investigated. This is the first phylogenetic tree for oribatid mites s.l. (incl. Astigmata) based on nucleotide sequences. Intraspecific genetic variation in the D3 region was very low, confirming the hypothesis that this region is a good species marker. Results from neighbour joining (NJ) and maximum parsimony (MP) algorithms indicate that several species-rich parthenogenetic groups like Camisiidae, Nanhermanniidae and Malaconothridae are monophyletic, consistent with the hypothesis that some oribatid mite groups diversified despite being parthenogenetic. The MP and maximum likelihood (ML) method indicated that the D3 region is a good tool for elucidating the relationship of oribatid mite species on a small scale(genera, families) but is not reliable for large-scale taxonomy, because branches from the NJ algorithm collapsed in the MP and ML tree. In all trees calculated by different algorithms the Astigmata clustered within the oribatid mites, as proposed earlier.

Animals↗

PubMed2001

Protein profile and biochemical properties of Bacillus circulans isolated from intestines of small free-living animals in Poland.

Forty-seven strains of Bacillus circulans isolated from the intestinal tract of free-living small mammals from the Narvia and Biebrza National Park (NE Poland) were compared with the electrophoretic whole-cell protein patterns on the basis of SDS-PAGE and biochemical characteristic using API tests (50 CHB and 20E). The strains were grouped into two clusters (I and II) at the similarity of protein pattern of 78% using the simple matching coefficient and clustering on unweighted pair group arithmetic average algorithm (SSM/UPGMA). Each of the clusters comprised reference strains. The rods included in the cluster I are characterized by higher biochemical activity in comparison with the bacilli II.

Animals↗

PubMed2001

Best harmony, unified RPCL and automated model selection for unsupervised and supervised learning on Gaussian mixtures, three-layer nets and ME-RBF-SVM models.

After introducing the fundamentals of BYY system and harmony learning, which has been developed in past several years as a unified statistical framework for parameter learning, regularization and model selection, we systematically discuss this BYY harmony learning on systems with discrete inner-representations. First, we shown that one special case leads to unsupervised learning on Gaussian mixture. We show how harmony learning not only leads us to the EM algorithm for maximum likelihood (ML) learning and the corresponding extended KMEAN algorithms for Mahalanobis clustering with criteria for selecting the number of Gaussians or clusters, but also provides us two new regularization techniques and a unified scheme that includes the previous rival penalized competitive learning (RPCL) as well as its various variants and extensions that performs model selection automatically during parameter learning. Moreover, as a by-product, we also get a new approach for determining a set of 'supporting vectors' for Parzen window density estimation. Second, we shown that other special cases lead to three typical supervised learning models with several new results. On three layer net, we get (i) a new regularized ML learning, (ii) a new criterion for selecting the number of hidden units, and (iii) a family of EM-like algorithms that combines harmony learning with new techniques of regularization. On the original and alternative models of mixture-of-expert (ME) as well as radial basis function (RBF) nets, we get not only a new type of criteria for selecting the number of experts or basis functions but also a new type of the EM-like algorithms that combines regularization techniques and RPCL learning for parameter learning with either least complexity nature on the original ME model or automated model selection on the alternative ME model and RBF nets. Moreover, all the results for the alternative ME model are also applied to other two popular nonparametric statistical approaches, namely kernel regression and supporting vector machine. Particularly, not only we get an easily implemented approach for determining the smoothing parameter in kernel regression, but also we get an alternative approach for deciding the set of supporting vectors in supporting vector machine.

Algorithms↗

PubMed2005

Morphological transition between diffusion-limited and ballistic aggregation growth patterns.

In this work, the transition between diffusion-limited (DLA) and ballistic aggregation (BA) models was reconsidered using a model in which biased random walks simulate the particle trajectories. The bias is controlled by a parameter lambda, which assumes the value lambda=0 (1) for the ballistic (diffusion-limited) aggregation model. Patterns growing from a single seed were considered. In order to simulate large clusters, an efficient algorithm was developed. For lambda (not equal to) 0 , the patterns are fractal on small length scales, but homogeneous on large ones. We evaluated the mean density of particles (-)rho in the region defined by a circle of radius r centered at the initial seed. As a function of r, (-)rho reaches the asymptotic value rho(0)(lambda) following a power law (-)rho = rho(0) +Ar(-gamma) with a universal exponent gamma=0.46 (2) , independent of lambda . The asymptotic value has the behavior rho(0) approximately |1-lambda|(beta) , where beta=0.26 (1) . The characteristic crossover length that determines the transition from DLA- to BA-like scaling regimes is given by xi approximately |1-lambda|(-nu) , where nu=0.61 (1) , while the cluster mass at the crossover follows a power law M(xi) approximately |1-lambda(-alpha) , where alpha=0.97 (2) . We deduce the scaling relations beta=nugamma and beta=2nu-alpha between these exponents.

Journal Article↗

PubMed1990

Gene conversion in the evolution of both the H-2 and Qa class I genes of the murine major histocompatibility complex.

In order to better understand the role of gene conversion in the evolution of the class I gene family of the major histocompatibility complex (MHC), we have used a computer algorithm to detect clustered sequence similarities among 24 class I DNA sequences from the H-2, Qa, and Tla regions of the murine MHC. Thirty-four statistically significant clusters were detected; individual analysis of the clusters suggested at least 25 past gene conversion or recombination events. These clusters are comparable in size to the conversions observed in the spontaneously occurring H-2Kbm and H-2Kkm2 mutations, and are distributed throughout all exons of the class I gene. Thus, gene conversion does not appear to be restricted to the regions of the class I gene encoding their antigen-presentation function. Moreover, both the highly polymorphic H-2 loci and the relatively monomorphic Qa and Tla loci appear to have participated as donors and recipients in conversion events. If gene conversion is not limited to the highly polymorphic loci of the MHC, then another factor, presumably natural selection, must be responsible for maintaining the observed differences in level of variation.

Algorithms↗

PubMed1994

Pulsatile fluctuations of plasma-gonadotropin-releasing hormone and corticotropin-releasing factor levels in healthy pregnant women.

Several data emphasize the neuroendocrine aspects of human placenta. Classical hypophyseotropic neurohormones are produced and secreted by the human placenta. Indeed, in contrast with non pregnant subjects, gonadotropin-releasing hormone (GnRH) and corticotropin-releasing factor (CRF) are measurable in maternal plasma during pregnancy. The aim of the present study was to investigate the characteristics of the secretory pattern of maternal plasma GnRH and CRF levels. A total of 25 healthy pregnant women participated in the present cross-sectional study. Plasma levels of the two neurohormones were measured according to three different time-intervals: 1) every five minutes for eight hours (n = 4), 2) every ten minutes for four hours (n = 15), 3) every four hours for 24 hours (n = 7). Following an acidic extraction plasma GnRH and CRF levels were measured by specific radioimmunoassay. Using two different algorithms (Detect and Cluster) a pulsatile pattern for both plasma GnRH and CRF levels was observed. Specific pulse frequency, amplitude, and duration were found. In the groups of women studied with a longer sampling interval, pulse frequency of GnRH or CRF levels did not differ between first and third trimester of gestation. However, amplitude of CRF pulses were significantly higher at term than at early stages of pregnancy (p < 0.01), while GnRH pulse amplitude was highest in women at first trimester of gestation. Evaluating the degree of concordance in all subjects, GnRH pulses significantly correlated with CRF pulses at 30 min. lag (p < 0.01). No significant circadian changes were found for any circulating neurohormone measured.(ABSTRACT TRUNCATED AT 250 WORDS)

Algorithms↗

PubMed2000

High-throughput functional annotation of novel gene products using document clustering.

Gene products differentially expressed in healthy vs. diseased tissues may be considered drug targets since the change in their expression level can be related to the cause and progression of the disease studied. A significant portion of the proteins produced by these genes will be unknown and consequently their function must be characterised. The experimental elucidation of biochemical function must be supported by computational tools which can help predicting the possible function of a given protein from its amino acid sequence. We have designed a high-throughput system which automatically analyses amino acid sequences deduced from differentially represented cDNA clones. The system attempts to assign a biological function to protein sequences by carrying out searches in sequence databanks and by locating functionally relevant motifs in the query sequences. The results delivered by the various prediction methods consist of the annotations of matching sequences and/or motifs, which are free-format texts written by humans and therefore may describe the same concept with synonymous words. It is desirable to present the results in such a way that the annotations describing the same biological function are grouped together. To this end we devised an algorithm that enables the hierarchical clustering of free-format documents based on their contents. The system is capable of detecting and flagging conflicting annotations, and will speed up the interpretation of the function prediction results.

Algorithms↗

PubMed2005

Combining smart darting with parallel tempering using Eckart space: application to Lennard-Jones clusters.

The smart-darting algorithm is a Monte Carlo based simulation method used to overcome quasiergodicity problems associated with disconnected regions of configurations space separated by high energy barriers. As originally implemented, the smart-darting method works well for clusters at low temperatures with the angular momentum restricted to zero and where there are no transitions to permutational isomers. If the rotational motion of the clusters is unrestricted or if permutational isomerization becomes important, the acceptance probability of darting moves in the original implementation of the method becomes vanishingly small. In this work the smart-darting algorithm is combined with the parallel tempering method in a manner where both rotational motion and permutational isomerization events are important. To enable the combination of parallel tempering with smart darting so that the smart-darting moves have a reasonable acceptance probability, the original algorithm is modified by using a restricted space for the smart-darting moves. The restricted space uses a body-fixed coordinate system first introduced by Eckart, and moves in this Eckart space are coupled with local moves in the full 3N-dimensional space. The modified smart-darting method is applied to the calculation of the heat capacity of a seven-atom Lennard-Jones cluster. The smart-darting moves yield significant improvement in the statistical fluctuations of the calculated heat capacity in the region of temperatures where the system isomerizes. When the modified smart-darting algorithm is combined with parallel tempering, the statistical fluctuations of the heat capacity of a seven-atom Lennard-Jones cluster using the combined method are smaller than parallel tempering when used alone.

Journal Article↗

PubMed2005

Learning from imbalanced data in surveillance of nosocomial infection.

OBJECTIVE: An important problem that arises in hospitals is the monitoring and detection of nosocomial or hospital acquired infections (NIs). This paper describes a retrospective analysis of a prevalence survey of NIs done in the Geneva University Hospital. Our goal is to identify patients with one or more NIs on the basis of clinical and other data collected during the survey. METHODS AND MATERIAL: Standard surveillance strategies are time-consuming and cannot be applied hospital-wide; alternative methods are required. In NI detection viewed as a classification task, the main difficulty resides in the significant imbalance between positive or infected (11%) and negative (89%) cases. To remedy class imbalance, we explore two distinct avenues: (1) a new re-sampling approach in which both over-sampling of rare positives and under-sampling of the noninfected majority rely on synthetic cases (prototypes) generated via class-specific sub-clustering, and (2) a support vector algorithm in which asymmetrical margins are tuned to improve recognition of rare positive cases. RESULTS AND CONCLUSION: Experiments have shown both approaches to be effective for the NI detection problem. Our novel re-sampling strategies perform remarkably better than classical random re-sampling. However, they are outperformed by asymmetrical soft margin support vector machines which attained a sensitivity rate of 92%, significantly better than the highest sensitivity (87%) obtained via prototype-based re-sampling.

Algorithms↗

PubMed2004

Diagnostic usefulness of PCR profiling of the differentially expressed marker genes in thyroid papillary carcinomas.

The study was set out to determine whether characteristic changes in the gene expression profile in papillary thyroid carcinoma (PTC) discovered by microarray assays can be used for conventional molecular diagnosis. Expression levels of five reported to be overexpressed and three underexpressed genes were examined in PTC and normal human tissues by real-time PCR and semi-quantitative duplex PCR. Stepwise logistic regression analysis, duplex PCR data evaluation with recursive partition machine algorithm and hierarchical cluster analysis identified SFTPB (upregulated) and TFF3 (downregulated) gene combination as most favorable for differential molecular diagnosis of PTC. Sensitivity, specificity and accuracy obtained in a series of histologically characterized thyroid tumor and normal tissue samples were 88.9, 96.7 and 94.9%, respectively. Applicability of the method to fine needle aspiration biopsy (FNAB) samples was demonstrated using a collection of needle washouts. In spite individual thyroid tumor and normal tissues as well as FNAB samples displayed a substantial degree of variability in the expression levels of analyzed genes, simultaneous molecular analysis of a panel of optimal markers allows making a high probability predictive estimate and may be considered as an informative method of preoperative PTC diagnosis.

Carcinoma, Papillary↗

PubMed2004

Tree analysis of mass spectral urine profiles discriminates transitional cell carcinoma of the bladder from noncancer patient.

BACKGROUND: Recent advances in proteomic profiling technologies, such as surface-enhanced laser desorption/ionization mass spectrometry (SELDI), have allowed preliminary profiling and identification of tumor markers in biological fluids in several cancer types and establishment of clinically useful diagnostic computational models. We developed a bioinformatics tool and used it to identify proteomic patterns in urine that distinguish transitional cell carcinoma (TCC) from noncancer. METHODS: Proteomic spectra were generated by mass spectroscopy (surface-enhanced laser desorption and ionization). A preliminary "training" set of spectra derived from analysis of urine from 46 TCC patients, 32 patients with benign urogenital diseases (BUD), and 40 age-matched unaffected healthy men were used to train and develop a decision tree classification algorithm that identified a fine-protein mass pattern that discriminated cancer from noncancer effectively. A blinded test set, including 38 new cases, was used to determine the sensitivity and specificity of the classification system. RESULTS: The algorithm identified a cluster pattern that, in the training set, segregated cancer from noncancer with sensitivity of 84.8% and specificity of 91.7%. The discriminatory pattern correctly identified. A sensitivity of 93.3% and a specificity of 87.0% for the blinded test were obtained when comparing the TCC vs. noncancer. CONCLUSIONS: These findings justify a prospective population-based assessment of proteomic pattern technology as a screening tool for bladder cancer in high-risk and general populations.

Adult↗

PubMed2005

Using wavelet de-noised spectra in NMR screening.

Principal component analysis (PCA) is a commonly used algorithm in multivariate analysis of NMR screening data. PCA substantially reduces the complexity of data in which a large number of variables are interrelated. For series of NMR spectra obtained for ligand binding, it is commonly used to visually group spectra with a similar response to ligand binding. A series of filters are applied to the experimental data to obtain suitable descriptors for PCA which optimize computational efficiency and minimize the weight of small chemical shift variations. The most common filter is bucketing where adjacent points are summed to a bucket. To overcome some inherent disadvantages of the bucketing procedure we have explored the effect of wavelet de-noising on multivariate analysis, using a series of HSQC spectra of proteins with different ligands present. The combination of wavelet de-noising and PCA is most efficient when PCA is applied to wavelet coefficients. This new algorithm yields good clustering and can be applied to series of one- or two-dimensional spectra.

Hydrogen↗

PubMed2002

Use of proteomic patterns in serum to identify ovarian cancer.

BACKGROUND: New technologies for the detection of early-stage ovarian cancer are urgently needed. Pathological changes within an organ might be reflected in proteomic patterns in serum. We developed a bioinformatics tool and used it to identify proteomic patterns in serum that distinguish neoplastic from non-neoplastic disease within the ovary. METHODS: Proteomic spectra were generated by mass spectroscopy (surface-enhanced laser desorption and ionisation). A preliminary "training" set of spectra derived from analysis of serum from 50 unaffected women and 50 patients with ovarian cancer were analysed by an iterative searching algorithm that identified a proteomic pattern that completely discriminated cancer from non-cancer. The discovered pattern was then used to classify an independent set of 116 masked serum samples: 50 from women with ovarian cancer, and 66 from unaffected women or those with non-malignant disorders. FINDINGS: The algorithm identified a cluster pattern that, in the training set, completely segregated cancer from non-cancer. The discriminatory pattern correctly identified all 50 ovarian cancer cases in the masked set, including all 18 stage I cases. Of the 66 cases of non-malignant disease, 63 were recognised as not cancer. This result yielded a sensitivity of 100% (95% CI 93--100), specificity of 95% (87--99), and positive predictive value of 94% (84--99). INTERPRETATION: These findings justify a prospective population-based assessment of proteomic pattern technology as a screening tool for all stages of ovarian cancer in high-risk and general populations.

CA-125 Antigen↗

PubMed2001

Protein--protein interaction maps: a lead towards cellular functions.

The availability of complete genome sequences now permits the development of tools for functional biology on a proteomic scale. Several experimental approaches or in silico algorithms aim at clustering proteins into networks with biological significance. Among those, the yeast two-hybrid system is the technology of choice to detect protein-protein interactions. Recently, optimized versions were applied at a genomic scale, leading to databases on the web. However, as with any other 'genetic' assay, yeast two-hybrid assays are prone to false positives and false negatives. Here we discuss these various technologies, their general limitations and the potential advances they make possible, especially when in combination with other functional genomics or bioinformatics analyses.

Animals↗

PubMed2005

Modeling the fractal growth of templated, mesoporous silica films.

Brewster angle micrographs have been obtained of the development of a silica film grown at an air-water interface with use of the surfactant template cetylpyridinium chloride. The micrograph images showed that the growing silica film exhibited complex, fractal-like patterns. These images have been analyzed to determine the values of the fractal dimension and lacunarity, and the forms of the autocorrelation function and Euclidean crossover behavior of the silica clusters have been observed. These statistical descriptors have been compared with the equivalent properties of simulated images of model structures generated by computer using a particular variant of the cluster-cluster aggregation (CCA) algorithm. Good agreement was found between the characteristic properties of typical experimental images and the simulated images. It was, therefore, suggested that the CCA process is a good model for the growth of the silica films.

Journal Article↗

PubMedDate not supplied

Estimation of brain compartment volume from MR Cavalieri slices.

PURPOSE: Recent theory has been developed to estimate volume from a systematic sample of tissue slices of a given thickness and to predict the corresponding error. Our goal was to check the error prediction formulas by resampling and to determine the minimum number of MR slices required to estimate the volumes of the cerebrum and of the compartments of gray matter (GM) and white matter (WM) with prescribed errors. METHOD: Our working data set comprised the GM and WM segmentations obtained from a paradigmatic high signal-to-noise ratio 3D spoiled GRASS MR volume data set for a single healthy human subject. The data were classified using a fuzzy clustering minimum distance algorithm. We thereby obtained a stack of 183 serial coronal slices of 1 mm thickness encompassing the whole cerebrum. Empirical resampling was carried out using the corresponding data vectors, and the theoretical error predictors were thereby checked for slice thicknesses of 1, 3, 9, and 27 mm, with a distance of 45 mm between slice midplanes. RESULTS: Irrespective of slice thickness, a minimum of 3, 5, and 10 slices provided estimates of the true total volume of GM and WM in the cerebrum with coefficients of error (CEs) of 10, 5, and 3%, respectively, where CE(V)% = 100 x SE(V)/V. For the cerebrum, a minimum of two, three, and four slices were required for CEs of the same precision. CONCLUSION: In combination with high signal-to-noise ratio and enhanced tissue contrast, Cavalieri slices are the most appropriate for MRI, they supply unbiased and highly efficient volume estimates of brain compartments. For a given number of slices, CE(V) decreases rapidly when the slices are thicker than the gaps between them; when the slices are thinner than the gaps, then CE(V) is similar to that in the situation when the slice thickness is zero.

Brain↗