Graph-based consensus clustering for class discovery from gene expression data

MOTIVATION: Consensus clustering, also known as cluster ensemble, is one of the important techniques for microarray data analysis, and is particularly useful for class discovery from microarray data. Compared with traditional clustering algorithms, consensus clustering approaches have the ability to integrate multiple partitions from different cluster solutions to improve the robustness, stability, scalability and parallelization of the clustering algorithms. By consensus clustering, one can discover the underlying classes of the samples in gene expression data. RESULTS: In addition to exploring a graph-based consensus clustering (GCC) algorithm to estimate the underlying classes of the samples in microarray data, we also design a new validation index to determine the number of classes in microarray data. To our knowledge, this is the first time in which GCC is applied to class discovery for microarray data. Given a pre specified maximum number of classes (denoted as K(max) in this article), our algorithm can discover the true number of classes for the samples in microarray data according to a new cluster validation index called the Modified Rand Index. Experiments on gene expression data indicate that our new algorithm can (i) outperform most of the existing algorithms, (ii) identify the number of classes correctly in real cancer datasets, and (iii) discover the classes of samples with biological meaning. AVAILABILITY: Matlab source code for the GCC algorithm is available upon request from Zhiwen Yu.     

d-Amino Acid-induced Transmural Potential Changes in Rat Small Intestine

The effect of d-amino acids on electrical potential across the wall of everted sacs of rat small intestine (ΔPD) was investigated. Valine, isoleucine, serine, threonine, histidine and arginine induced ΔPD of decreasing the mucosal negativity, which was largely responsible to streaming potential. On the contrary, tryptophan, leucine, phenylalanine, alanine, lysine and methionine induced positive ΔPD, the results being implicit of an active process for their transport.

d-Leucine-induced ΔPD depended on the concentration of both the amino acid and Na⁺. and followed saturation kinetics, the features being in support of its active transfer. Kinetic parameters calculated from ΔPD differed significantly by the isomer of leucine, suggesting that d-preferring system exists besides l-preferring one, or alternatively that mechanism of ΔPD induction by d-Ieucine differs, at least in part, from that by l-leucine.     

Plasmid-mediated nicotine degradation inArthrobacter oxidans

The spontaneous loss byArthrobacter oxidans cells of the nicotine-degrading ability (Nic⁺) was 0.06%. It could be increased by treatment with plasmid-curing agents up to 8%. It was possible by conjugation to restore the Nic⁺ phenotype in such cured derivatives and to transfer the Nic⁺ character to Nic^- Arthrobacter species. Plasmid DNA, 160 kb in size as judged by contour length measurements, could be isolated from cleared lysates ofA. oxidans cells by acridine yellow chromatography. Agarose gel electrophoresis of DNA isolated fromArthrobacter exconjugates revealed the occurrence of plasmid DNA within these strains; its mobility was similar to that of the plasmid DNA present inA. oxidans. Although the expression and inducibility of the transferred genes was poor in most of theArthrobacter species exconjugants, apparently authentic 6-hydroxy-l-nicotine oxidase could be identified in these cells after enrichment by an enzyme-specific chromatography.Abbreviations 6-HDNO 6-hydroxy-d-nicotine oxidase - 6-HLNO 6-hydroxy-l-nicotine oxidase - kb kilobase - Nic⁺ ability to usel- ord-nicotine as sole carbon and nitrogen source - Nic^- absence of Nic⁺ character Enzymes (EC 1.5.3.5.) 6-Hydroxy-l-nicotine oxidase, 6-hydroxy-l-nicotine: oxygen oxidoreductase - (EC 1.5.3.6.) 6-hydroxy-d-nicotine oxidase, 6-hydroxy-d-nicotine: oxygen oxidoreductase - (EC 3.1.4.22) ribonuclease A, ribonucleate 3-pyrimidino-oligo-nucleotidohydrolase     

Engineering of pentose transport in Corynebacterium glutamicum to improve simultaneous utilization of mixed sugars

Corynebacterium glutamicum strains CRA1 and CRX2 are able to grow on l-arabinose and d-xylose, respectively, as sole carbon sources. Nevertheless, they exhibit the major shortcoming that their sugar consumption appreciably declines at lower concentrations of these substrates. To address this, the C. glutamicum ATCC31831 l-arabinose transporter gene, araE, was independently integrated into both strains. Unlike its parental strain, resultant CRA1-araE was able to aerobically grow at low (3.6 g·l⁻¹) l-arabinose concentrations. Interestingly, strain CRX2-araE grew 2.9-fold faster than parental CRX2 at low (3.6 g·l⁻¹) d-xylose concentrations. The corresponding substrate consumption rates of CRA1-araE and CRX2-araE under oxygen-deprived conditions were 2.8- and 2.7-fold, respectively, higher than those of their respective parental strains. Moreover, CRA1-araE and CRX2-araE utilized their respective substrates simultaneously with d-glucose under both aerobic and oxygen-deprived conditions. Based on these observations, a platform strain, ACX-araE, for C. glutamicum-based mixed sugar utilization was designed. It harbored araBAD for l-arabinose metabolism, xylAB for d-xylose metabolism, d-cellobiose permease-encoding bglF ^317A , β-glucosidase-encoding bglA and araE in its chromosomal DNA. In mineral medium containing a sugar mixture of d-glucose, d-xylose, l-arabinose, and d-cellobiose under oxygen-deprived conditions, strain ACX-araE simultaneously and completely consumed all sugars.     

An improved algorithm for the detection of genomic variation using short oligonucleotide expression microarrays

High‐throughput microarray experiments often generate far more biological information than is required to test the experimental hypotheses. Many microarray analyses are considered finished after differential expression and additional analyses are typically not performed, leaving untapped biological information left undiscovered. This is especially true if the microarray experiment is from an ecological study of multiple populations. Comparisons across populations may also contain important genomic polymorphisms, and a subset of these polymorphisms may be identified with microarrays using techniques for the detection of single feature polymorphisms (SFP). SFPs are differences in microarray probe level intensities caused by genetic polymorphisms such as single‐nucleotide polymorphisms and small insertions/deletions and not expression differences. In this study, we provide a new algorithm for the detection of SFPs, evaluate the algorithm using existing data from two publicly available Affymetrix Barley (Hordeum vulgare) microarray data sets and compare them to two previously published SFP detection algorithms. Results show that our algorithm provides more consistent and sensitive calling of SFPs with a lower false discovery rate. Simultaneous analysis of SFPs and differential expression is a low‐cost method for the enhanced analysis of microarray data, enabling additional biological inferences to be made.     

Horizontal Gene Transfer Involved in the Convergent Evolution of the Plasmid-Encoded Enantioselective 6-Hydroxynicotine Oxidases

The D- and L-specific nicotine oxidases are flavoproteins involved in the oxidative degradation of nicotine by the Gram-positive soil bacterium Arthrobacter nicotinovorans. Their structural genes are located on a 160-kbp plasmid together with those of other nicotine-degrading enzymes. They are structurally unrelated at the DNA as well as at the protein level. Each of these oxidases possesses a high degree of substrate specificity; their catalytic stereoselectivity is absolute, although they are able to bind both enantiomeric substrates with a similar affinity. It appears that the existence of these enzymes is the result of convergent evolution. The amino acid sequence of 6-hydroxy-l-nicotine oxidase (EC 1.5.3.6) as derived from the respective structural gene shows considerable structural similarity with eukaryotic monoamine oxidases (EC 1.4.3.4) but not with monoamine oxidases from prokaryotic bacteria including those of the genus Arthrobacter. These similarities are not confined to the nucleotide-binding sites. A 100-amino acid stretch at the N-terminal regions of 6-hydroxy-l-nicotine oxidase and human monoamine oxidases A possess a 35% homology. Overall, 27.0, 26.9, and 25.8% of the amino acid positions of the monoamine oxidases of Aspergillus niger (N), humans (A), and rainbow trout (Salmo gairdneri) are identical to those of 6-hydroxy-l-nicotine oxidase (Smith–Waterman algorithm). In addition, the G+C content of the latter enzyme is in the range of that of eukaryotic monoamine oxidases and definitely lower than that of the A. nicotinovorans DNA and even that of the pAO1 DNA. The primary structure of 6-hydroxy-d-nicotine oxidase (EC 1.5.3.5) does not reveal its evolutionary history as easily. Significant similarities are found with a mitomycin radical oxidase from Streptomyces lavendulae (23.3%) and a ``hypothetical protein' from Mycobacterium tuberculosis (26.0%). It is proposed that the plasmid-encoded gene of 6-hydroxy-l-nicotine oxidase evolved after horizontal transfer from an eukaryotic source. Received: 6 March 1998 / Accepted: 15 July 1998     

Screening for candidate genes involved in tolerance to organic solvents in yeast

Saccharomyces cerevisiae mutant strain, KK-211, isolated from serial culture in medium containing isooctane showed an extremely higher tolerance to the hydrophobic organic-solvents, which are toxic to yeast cells compared to the wild-type parent strain, DY-1. To detect genes that are related to this tolerance, a DNA microarray analysis was performed using mRNAs isolated from strains DY-1 and KK-211. Fourteen genes were identified as being related to the tolerance. The expression of 12 genes including ICT1, YNL190W, and PRY3, was induced while the expression of two genes including PHO84 was repressed in strain KK-211. Two genes, ICT1 and YNL190W showed the same profile in the DNA microarray analysis and a differential display-polymerase chain reaction analysis. But, there is no detectable difference in the expression profile of KK-211 cells cultured with or without isooctane. The results suggest that change in expression levels of multiple genes that confer the modification function of the cell surface, not by a single gene, might be required for yeast cell tolerance to organic solvents.     

Use of a polymerase-chain-reaction-amplified DNA probe from Pseudomonas putida to detect d-hydantoinase-producing microorganisms by direct colony hybridization

Pseudomonas putida strain DSM 84 produces N-carbamyl-d-amino acids from the corresponding d-5-monosubstituted hydantoins. The sequence of the d-hydantoinase gene from this strain (GenBank accession number L24157) was used to develop a DNA probe of 122 base pairs (bp) that could detect d-hydantoinase genes in other bacterial genera by DNA and by colony hybridization. Under conditions tolerating 32% mismatch, the probe was specific for all strains that expressed d-hydantoinase activity. These include Pseudomonadaceae of all rRNA groups, and bacteria belonging to the genera Agrobacterium, Serratia, Corynebacterium, and Arthrobacter. Environmental sampling was simulated by screening a mixture of unknown microorganims from commercial inocula for the biodegradation of industrial, municipal and domestic wastes. The 122-bp probe was specific for microorganisms that subsequently demonstrated d-hydantoinase activity. Bacterial species from four different genera were detected, which were Pseudomonas, Klebsiella, Enterobacter, and Enterococcus.     

Search for signatures in miRNAs associated with cancer

Since the first discovery in the early 1990''s, the predicted and validated population of microRNAs (miRNAs or miRs) has grown significantly. These small (~22 nucleotides long) regulators of gene expression have been implicated and associated with several genes in the cancer pathway as well. Globally, the identification and verification of microRNAs as biomarkers for cancer cell types has been the area of thrust for most miRNA biologists. However, there has been a noticeable vacuum when it comes to identifying a common signature or trademark that could be used to demarcate a miR to be associated with the development or suppression of cancer. To answer these queries, we report an in silico study involving the identification of global signatures in experimentally validated microRNAs which have been associated with cancer. This study has thrown light on the presence of significant common signatures, viz., - sequential and hybridization, which may distinguish a miR to be associated with cancer. Based on our analysis, we suggest the utility of such signatures in the design and development of algorithms for prediction of miRs involved in the cancer pathway.     

Characterization of prolidase I and II purified from normal human erythrocytes: comparison with prolidase in erythrocytes from a patient with prolidase deficiency

The effect of various sulfur-containing amino acids on the activities of prolidase isoenzymes I and II isolated from erythrocytes of healthy individuals, and erythrocyte lysates from a patient with prolidase deficiency was investigated. The activity of prolidase I against glycylproline was strongly enhanced by d-methionine. l-Methionine and d,l-methionine slightly enhanced the activity at low concentration, but N-acetyl-l-methionine had no effect. d-Ethionine, l-ethionine, and d,l-ethionine also enhanced the activity of prolidase I. d,l-Homocysteine enhanced the activity at low concentration, but inhibited the activity at 50 mM. The activity of prolidase II against methionylproline was enhanced by d-methionine, d,l-methionine, and l-methionine, but N-acetyl-l-methionine had no effect. d-Ethionine and d,l-ethionine strongly enhanced the activity of prolidase II compared with l-ethionine; d,l-homocysteine weakly enhanced the activity. d,l-Homocysteine-thiolactone inhibited the activities of prolidase I and II in a concentration-dependent manner. The effect of various sulfur-containing amino acids on prolidase activity against methionylproline in erythrocyte lysates from a patient with prolidase deficiency was almost the same as that on prolidase II. The kinetics of the activities of prolidase I, II, and patient prolidase were also studied. Their K _m values were changed by adding sulfur-containing amino acids, but V _max values were unchanged.     

Influence of surface penetration on measured fluid force on a hand model

The purpose of this study was to quantify the effect of wave drag due to surface penetration on drag and lift forces (C_d and C_l) acting on a hand model. The values of C_d and C_l had been acquired to gain the hydrodynamic characteristics of the swimmer's hand and predict force on the swimmer's hand. These values have also been used to benchmark computational fluid dynamics analysis. Because the previous studies used a hand/forearm model which penetrated the water's surface, the values of C_d and C_l include the effect of the surface wave on the model. Wave formation causes pressure differences between the frontal and rear sides of a surface-penetrating model as a result of depressions and elevations in the water's surface. This may be considered as wave drag due to surface penetration. Fluid forces due to wave drag on the forearm should not be included in the measured C_d and C_l of a swimmer's hand that does not sweep near the water's surface. Two hand/forearm models are compared, one with the hand rigidly connected to the forearm. The other model was constructed to isolate the fluid forces acting on the hand from the influence of wave drag on the forearm. The measurements showed that the effect of wave drag on the hand model caused large increases in the values of C_d, up to 46–98% with lesser increases in C_l of 2–12% depending on the hand orientation. The present study provides an improved method to determine the values of C_d and C_l that eliminates the effect of wave drag on a hand/forearm model by isolating the measurement of fluid forces on the forearm of the hand/forearm model in order to separately acquire the forces on the hand.     

Removal of <Emphasis Type="SmallCaps">l</Emphasis>-alanine from the production of <Emphasis Type="SmallCaps">l</Emphasis>-2-aminobutyric acid by introduction of alanine racemase and <Emphasis Type="SmallCaps">d</Emphasis>-amino acid oxidase

l-2-Aminobutyric acid can be synthesized in a transamination reaction from l-threonine and l-aspartic acid as substrates by the action of threonine deaminase and aromatic aminotransferase, but the by-product l-alanine was produced simultaneously. A small amount of l-alanine increased the complexity of the l-2-aminobutyric acid recovery process because of their extreme similarity in physical and chemical properties. Acetolactate synthase has been introduced to remove the pyruvate intermediate for reducing the l-alanine concentration partially. To eliminate the remnant l-alanine, alanine racemase of Bacillus subtilis in combination with d-amino acid oxidase of Rhodotorula gracilis or Trigonopsis variabilis respectively was introduced into the reaction system for the l-2-aminobutyric acid synthesis. l-Alanine could be completely removed by the action of alanine racemase of B. subtilis and d-amino acid oxidase of R. gracilis; thereby, high-purity l-2-aminobutyric acid was achieved. The results revealed that alanine racemase could discriminate effectively between l-alanine and l-2-aminobutyric acid, and selectively catalyzed l-alanine to d-alanine reversibly. d-Amino acid oxidase then catalyzed d-alanine to pyruvate stereoselectively. Furthermore, this method was also successfully used to remove the by-product l-alanine in the production of other neutral amino acids such as l-tertiary leucine and l-valine, suggesting that multienzymatic whole-cell catalysis can be employed to provide high purity products.     

The Production of Xylitol,l-Arabinitol and Ribitol by Yeasts

The polyalcohol production from the pentoses such as d-xylose, l-arabinose and d-ribose by various genera and species of yeasts was examined. Candida polymorpha dissimilated aerobically these three pentoses and produced xylitol from d-xylose, l-arabinitol from l-arabinose and ribitol from d-ribose at good yield of 30~40% of sugar consumed. The result suggests that these polyalcohols would be major products from pentoses by yeasts, but some unidentified minor polyalcohols were also produced.     

A comprehensive comparison of random forests and support vector machines for microarray-based cancer classification

Background  

Cancer diagnosis and clinical outcome prediction are among the most important emerging applications of gene expression microarray technology with several molecular signatures on their way toward clinical deployment. Use of the most accurate classification algorithms available for microarray gene expression data is a critical ingredient in order to develop the best possible molecular signatures for patient care. As suggested by a large body of literature to date, support vector machines can be considered "best of class" algorithms for classification of such data. Recent work, however, suggests that random forest classifiers may outperform support vector machines in this domain.     

Microbial Resolution of Racemic Carvomenthols

By selected microorganisms dl-carvomenthyl acetate, dl-isocarvomenthyl acetate and dl-neo isocarvomenthyl acetate were asymmetrically hydrolyzed to l-carvomenthol with d-carvomenthyl acetate, l-isocarvomenthol with d-isocarvomenthyl acetate and d-neo isocarvomenthol with l-neo isocarvomenthyl acetate respectively; dl-neo carvomenthyl acetate was not hydrolyzed.     

Single Taste Stimuli Elicit Either Increases or Decreases in Intracellular Calcium in Isolated Catfish Taste Cells

Taste cells are specialized epithelial cells that respond to stimulation with release of neurotransmitters onto afferent nerves that innervate taste buds. In analogy to neurotransmitter release in other cells, it is expected that neurotransmitter release in taste cells is dependent on an increase in intracellular Ca²⁺ ([Ca²⁺] _i ). We have studied changes in [Ca²⁺] _i elicited by the taste stimuli l- and d-arginine in isolated taste cells from the channel catfish (Ictalurus punctatus). In a sample of 119 cells, we found 15 cells responding to l-arginine, and 12 cells responding to d-arginine with an increase in [Ca²⁺] _i . The response to l-arginine was inhibited by equimolar d-arginine in cells where d-arginine alone did not cause a change in [Ca²⁺] _i , which is consistent with mediation of this response by a previously characterized l-arginine-gated nonspecific cation channel antagonized by d-arginine [31]. However, we also found that these taste stimuli elicited decreases in [Ca²⁺] _i in substantial number of cells (6 for l-Arg, and 2 for d-Arg, n= 119). These observations suggest that stimulation of taste cells with sapid stimuli may result in simultaneous excitation and inhibition of different taste cells within the taste bud, which could be involved in local processing of the taste signal. Received: 25 May 1995/Revised: 29 September 1995     

Anaerobic growth of Salmonella typhimurium on l(+)- and d(−)-tartrate involves an oxaloacetate decarboxylase Na+ pump

We show here that the Enterobacterium Salmonella typhimurium LT2 has the capacity to grow anaerobically on l(+)- or d(-)-tartrate as sole carbon and energy source. Growth on these substrates was Na⁺-dependent and involved the l(+)- or d(-)-tartrate-inducible expression of oxaloacetate decarboxylase. The induced decarboxylase was closely related to the oxaloacetate decarboxylase Na⁺ pump of Klebsiella pneumoniae as shown by the sensitivity towards avidin, the location in the cytoplasmic membrane, activation by Na⁺ ions, and Western blot analysis with antiserum raised against the K. pneumoniae oxaloacetate decarboxylase. Participation of an oxaloacetate decarboxylase Na⁺ pump in l(+)-tartrate degradation by S. typhimurium is in accord with results from DNA analyses. The deduced protein sequence of the open reading frame identified upstream of the recently sequenced oxaloacetate decarboxylase genes is clearly homologous with the -subunit of l-tartrate dehydratase from Escherichia coli. Southern blot analysis with S. typhimurium chromosomal DNA indicated the presence of probably more than one gene for oxaloacetate decarboxylase.