Constructing localized non-linear projection manifolds in a principled way:hierarchical generative topographic mapping

It has been argued that a single two-dimensional visualization plot may not be sufficient to capture all of the interesting aspects of complex data sets, and therefore a hierarchical visualization system is desirable. In this paper we extend an existing locally linear hierarchical visualization system PhiVis (Bishop98a) in several directions: 1. We allow for em non-linear projection manifolds. The basic building block is the Generative Topographic Mapping. 2. We introduce a general formulation of hierarchical probabilistic models consisting of local probabilistic models organized in a hierarchical tree. General training equations are derived, regardless of the position of the model in the tree. 3. Using tools from differential geometry we derive expressions for local directionalcurvatures of the projection manifold. Like PhiVis, our system is statistically principled and is built interactively in a top-down fashion using the EM algorithm. It enables the user to interactively highlight those data in the parent visualization plot which are captured by a child model.We also incorporate into our system a hierarchical, locally selective representation of magnification factors and directional curvatures of the projection manifolds. Such information is important for further refinement of the hierarchical visualization plot, as well as for controlling the amount of regularization imposed on the local models. We demonstrate the principle of the approach on a toy data set andapply our system to two more complex 12- and 19-dimensional data sets.

Visual data mining: integrating machine learning with information visualization

Today, the data available to tackle many scientific challenges is vast in quantity and diverse in nature. The exploration of heterogeneous information spaces requires suitable mining algorithms as well as effective visual interfaces. Most existing systems concentrate either on mining algorithms or on visualization techniques. Though visual methods developed in information visualization have been helpful, for improved understanding of a complex large high-dimensional dataset, there is a need for an effective projection of such a dataset onto a lower-dimension (2D or 3D) manifold. This paper introduces a flexible visual data mining framework which combines advanced projection algorithms developed in the machine learning domain and visual techniques developed in the information visualization domain. The framework follows Shneiderman’s mantra to provide an effective user interface. The advantage of such an interface is that the user is directly involved in the data mining process. We integrate principled projection methods, such as Generative Topographic Mapping (GTM) and Hierarchical GTM (HGTM), with powerful visual techniques, such as magnification factors, directional curvatures, parallel coordinates, billboarding, and user interaction facilities, to provide an integrated visual data mining framework. Results on a real life high-dimensional dataset from the chemoinformatics domain are also reported and discussed. Projection results of GTM are analytically compared with the projection results from other traditional projection methods, and it is also shown that the HGTM algorithm provides additional value for large datasets. The computational complexity of these algorithms is discussed to demonstrate their suitability for the visual data mining framework.

Visual data mining using principled projection algorithms and information visualization techniques

We introduce a flexible visual data mining framework which combines advanced projection algorithms from the machine learning domain and visual techniques developed in the information visualization domain. The advantage of such an interface is that the user is directly involved in the data mining process. We integrate principled projection algorithms, such as generative topographic mapping (GTM) and hierarchical GTM (HGTM), with powerful visual techniques, such as magnification factors, directional curvatures, parallel coordinates and billboarding, to provide a visual data mining framework. Results on a real-life chemoinformatics dataset using GTM are promising and have been analytically compared with the results from the traditional projection methods. It is also shown that the HGTM algorithm provides additional value for large datasets. The computational complexity of these algorithms is discussed to demonstrate their suitability for the visual data mining framework. Copyright 2006 ACM.

Semisupervised learning of hierarchical latent trait models for data visualization

Recently, we have developed the hierarchical Generative Topographic Mapping (HGTM), an interactive method for visualization of large high-dimensional real-valued data sets. In this paper, we propose a more general visualization system by extending HGTM in three ways, which allows the user to visualize a wider range of data sets and better support the model development process. 1) We integrate HGTM with noise models from the exponential family of distributions. The basic building block is the Latent Trait Model (LTM). This enables us to visualize data of inherently discrete nature, e.g., collections of documents, in a hierarchical manner. 2) We give the user a choice of initializing the child plots of the current plot in either interactive, or automatic mode. In the interactive mode, the user selects "regions of interest," whereas in the automatic mode, an unsupervised minimum message length (MML)-inspired construction of a mixture of LTMs is employed. The unsupervised construction is particularly useful when high-level plots are covered with dense clusters of highly overlapping data projections, making it difficult to use the interactive mode. Such a situation often arises when visualizing large data sets. 3) We derive general formulas for magnification factors in latent trait models. Magnification factors are a useful tool to improve our understanding of the visualization plots, since they can highlight the boundaries between data clusters. We illustrate our approach on a toy example and evaluate it on three more complex real data sets. © 2005 IEEE.

Risk factors for ocular surface squamous neoplasia recurrence after treatment with topical mitomycin C and interferon alpha-2b

Purpose To determine the rate of recurrence and associated risk factors following the use of mitomycin C (MMC) and/or interferon alpha-2b (IFN) for management of non-invasive ocular surface squamous neoplasia (OSSN). Design Retrospective non-comparative interventional case series. Methods Clinical practice setting of 135 patients treated consecutively with topical MMC (0.4 mg/mL) and/or IFN (1 million units/mL) for OSSN observed for clinical recurrence. Results Clinical recurrences were diagnosed in 19 of 135 (14.1%) eyes following topical treatment. The mean time to recurrence was 17.2 months (range 4 - 61) with 14 (73.7%) recurring within a two year period. There was no greater risk of recurrence identified for variables including lesion size, lesion location, gender, age, treatment type or duration. Post-hoc log-Rank pairwise comparisons revealed that lesions initially treated using surgery alone had significantly reduced time to recurrence (21.1 ± 5.6 months) compared to previous topical treatment with MMC (with or without surgery) (29.6 ± 4.7 months) (p = 0.04) and primary OSSN (23.2 ± 1.8 months) (p = 0.09). Conclusions Topical MMC and IFN are an effective treatment modality for a wide range of non-invasive OSSN. Topical therapy avoids the morbidity of excisional surgery with equivalent or reduced recurrence rates and should be considered as primary therapy.

Evaluation of estrogen receptor alpha and beta and progesterone receptor expression and correlation with clinicopathologic factors and proliferative marker Ki-67 in breast cancers

To elucidate the molecular profile of hormonal steroid receptor status, we analyzed ER-alpha, ER-beta, and PGR mRNA and protein expression in 80 breast carcinomas using reverse transcriptase polymerase chain reaction (RT-PCR), quantitative RT-PCR, and immunohistochemical analysis. Qualitative analysis revealed positive expression of ER-alpha, ER-beta, and PGR mRNA in 48%, 59%, and 48% of the breast carcinomas, respectively. ER-alpha, ER-beta, and PGR transcript overexpression was observed in 51%, 0%, and 12% of the cases, respectively, whereas moderate or strong protein expression was detected in 68%, 78%, and 49% of the cases, respectively. Tumor grade was negatively correlated with transcript and protein levels of ER-alpha (P = .0169 and P = .0006, respectively) and PGR (P = .0034 and P = .0005, respectively). Similarly, proliferative index Ki-67 was negatively associated with transcript and protein levels of ER-alpha (P = .0006 and P < .0001, respectively) and PGR (P = .0258 and P =. 0005, respectively). These findings suggest that ER-alpha and PGR expression are associated with well-differentiated breast tumors and less directly related to cell proliferation. A significant statistical difference was observed between lymph node status and ER-beta protein expression (P = .0208). In ER-alpha-negative tumors, we detected a correlation between ER-beta protein expression and high levels of Ki-67. These data suggest that ER-beta could be a prognostic marker in human breast cancer. (C) 2008 Elsevier B.V. All rights reserved.

Rheological characterisation and effect of abiotic factors on the antimicrobial efficacy of chitosan-based hidrogels containing alpha-hydroxy acids

The aim of this study was to investigate the rheological properties and antibacterial efficacy of chitosan/ alpha-hydroxy acids (lactic acid and glycolic acid) and cellulose polymers, both in hydrogels, in order to produce a formulation with improved activity against Propionibacterium acnes and Staphylococcus aureus, which can potentially be used in the treatment of acne. The rheological characterisation of the hydrogels was examined using continuous shear and viscoelastic creep. The antibacterial activities of formulations were performed by the well diffusion and broth microdilution. The hydrogels formulated with only chitosan showed pseudoplastic behavior while the chitosan hydrogels with cellulose polymers presented viscoelastic properties. The antibacterial activity was proportional to AHA and chitosan concentration. It was enhanced at low pH values and with high molecular weight chitosan and did not change with the incorporation of two cellulose polymers. The antibacterial mechanism of chitosan has currently been hypothesized as being related to surface interference. The results show that chitosan - based hydrogels containing AHA and cellulose polymers are viscoelastic,indicating good applicability onto the skin, and they present bacterial activity under various experimental conditions.

Positioning of two alpha subunit carboxy-terminal domains of RNA polymerase at promoters by two transcription factors

Interactions between the cAMP receptor protein (CRP) and the carboxy-terminal regulatory domain (CTD) of Escherichia coli RNA polymerase α subunit were analyzed at promoters carrying tandem DNA sites for CRP binding using a chemical nuclease covalently attached to α. Each CRP dimer was found to direct the positioning of one of the two α subunit CTDs. Thus, the function of RNA polymerase may be subject to regulation through protein–protein interactions between the two α subunits and two different species of transcription factors.

Transdifferentiation of myoblasts by the adipogenic transcription factors PPAR gamma and C/EBP alpha.

Skeletal muscle and adipose tissue development often has a reciprocal relationship in vivo, particularly in myodystrophic states. We have investigated whether determined myoblasts with no inherent adipogenic potential can be induced to transdifferentiate into mature adipocytes by the ectopic expression of two adipogenic transcription factors, PPAR gamma and C/EBP alpha. When cultured under optimal conditions for muscle differentiation, murine G8 myoblasts expressing PPAR gamma and C/EBP alpha show markedly reduced levels of the myogenic basic helix-loop-helix proteins MyoD, myogenin, MRF4, and myf5 and are completely unable to differentiate into myotubes. Under conditions permissive for adipogenesis including a PPAR activator, these cells differentiate into mature adipocytes that express molecular markers characteristic of this lineage. Our results demonstrate that a developmental switch between these two related but highly specialized cell types can be controlled by the expression of key adipogenic transcription factors. These factors have an ability to inhibit myogenesis that is temporally and functionally separate from their ability to stimulate adipogenesis.

Tumor necrosis factors alpha and beta protect neurons against amyloid beta-peptide toxicity: evidence for involvement of a kappa B-binding factor and attenuation of peroxide and Ca2+ accumulation.

In Alzheimer disease (AD) the amyloid beta-peptide (A beta) accumulates in plaques in the brain. A beta can be neurotoxic by a mechanism involving induction of reactive oxygen species (ROS) and elevation of intracellular free calcium levels ([Ca2+]i). In light of evidence for an inflammatory response in the brain in AD and reports of increased levels of tumor necrosis factor (TNF) in AD brain we tested the hypothesis that TNFs affect neuronal vulnerability to A beta. A beta-(25-35) and A beta-(1-40) induced neuronal degeneration in a concentration- and time-dependent manner. Pretreatment of cultures for 24 hr with TNF-beta or TNF-alpha resulted in significant attenuation of A beta-induced neuronal degeneration. Accumulation of peroxides induced in neurons by A beta was significantly attenuated in TNF-pretreated cultures, and TNFs protected neurons against iron toxicity, suggesting that TNFs induce antioxidant pathways. The [Ca2+]i response to glutamate (quantified by fura-2 imaging) was markedly potentiated in neurons exposed to A beta, and this action of A beta was suppressed in cultures pretreated with TNFs. Electrophoretic mobility-shift assays demonstrated an induction of a kappa beta-binding activity in hippocampal cells exposed to TNFs. Exposure of cultures to I kappa B (MAD3) antisense oligonucleotides, a manipulation designed to induce NF-kappa B, mimicked the protection by TNFs. These data suggest that TNFs protect hippocampal neurons against A beta toxicity by suppressing accumulation of ROS and Ca2+ and that kappa B-dependent transcription is sufficient to mediate these effects. A modulatory role for TNF in the neurodegenerative process in AD is proposed.

Experimental data of CO2, CO3, Omega calcite saturation, pH values, alpha and epsilon fractionation factors, and Dd18O calctite-water

The oxygen isotopic composition (d18O) of calcium carbonate of planktonic calcifying organisms is a key tool for reconstructing both past seawater temperature and salinity. The calibration of paloeceanographic proxies relies in general on empirical relationships derived from field experiments on extant species. Laboratory experiments have more often than not revealed that variables other than the target parameter influence the proxy signal, which makes proxy calibration a challenging task. Understanding these secondary or "vital" effects is crucial for increasing proxy accuracy. We present data from laboratory experiments showing that oxygen isotope fractionation during calcification in the coccolithophore Calcidiscus leptoporus and the calcareous dinoflagellate Thoracosphaera heimii is dependent on carbonate chemistry of seawater in addition to its dependence on temperature. A similar result has previously been reported for planktonic foraminifera, supporting the idea that the [CO3]2- effect on d18O is universal for unicellular calcifying planktonic organisms. The slopes of the d18O/[CO3]2- relationships range between -0.0243 per mil/(µmol/kg) (calcareous dinoflagellate T. heimii) and the previously published -0.0022 per mil/(µmol/kg) (non-symbiotic planktonic foramifera Orbulina universa), while C. leptoporus has a slope of -0.0048 per mil/(µmol/kg). We present a simple conceptual model, based on the contribution of d18O-enriched [HCO3]- to the [CO3]2- pool in the calcifying vesicle, which can explain the [CO3]2- effect on d18O for the different unicellular calcifiers. This approach provides a new insight into biological fractionation in calcifying organisms. The large range in d18O/[CO3]2- slopes should possibly be explored as a means for paleoreconstruction of surface [CO3]2-, particularly through comparison of the response in ecologically similar planktonic organisms.

Measuring factors that influence the utilisation of preventive care services provided by general practitioners in Australia

Background: Relatively little research attention has been given to the development of standardised and psychometrically sound scales for measuring influences relevant to the utilisation of health services. This study aims to describe the development, validation and internal reliability of some existing and new scales to measure factors that are likely to influence utilisation of preventive care services provided by general practitioners in Australia.----- Methods: Relevant domains of influence were first identified from a literature review and formative research. Items were then generated by using and adapting previously developed scales and published findings from these. The new items and scales were pre-tested and qualitative feedback was obtained from a convenience sample of citizens from the community and a panel of experts. Principal Components Analyses (PCA) and internal reliability testing (Cronbach's alpha) were then conducted for all of the newly adapted or developed scales utilising data collected from a self-administered mailed survey sent to a randomly selected population-based sample of 381 individuals (response rate 65.6 per cent).----- Results: The PCA identified five scales with acceptable levels of internal consistency were: (1) social support (ten items), alpha 0.86; (2) perceived interpersonal care (five items), alpha 0.87, (3) concerns about availability of health care and accessibility to health care (eight items), alpha 0.80, (4) value of good health (five items), alpha 0.79, and (5) attitudes towards health care (three items), alpha 0.75.----- Conclusion The five scales are suitable for further development and more widespread use in research aimed at understanding the determinants of preventive health services utilisation among adults in the general population.

Akt, AS160, metabolic risk factors and aerobic fitness in middle-aged women

Aims: This study investigated the association between the basal (rest) insulin-signaling proteins, Akt, and the Akt substrate AS160, metabolic risk factors, inflammatory markers and aerobic fitness, in middle-aged women with varying numbers of metabolic risk factors for type 2 diabetes. Methods: Sixteen women (n = 16) aged 51.3+/-5.1 (mean +/-SD) years provided muscle biopsies and blood samples at rest. In addition, anthropometric characteristics and aerobic power were assessed and the number of metabolic risk factors for each participant was determined (IDF criteria). Results: The mean number of metabolic risk factors was 1.6+/-1.2. Total Akt was negatively correlated with IL-1 beta (r = -0.45, p = 0.046), IL-6 (r = -0.44, p = 0.052) and TNF-alpha (r = -0.51, p = 0.025). Phosphorylated AS160 was positively correlated with HDL (r = 0.58, p = 0.024) and aerobic fitness (r = 0.51, p = 0.047). Furthermore, a multiple regression analysis revealed that both HDL (t = 2.5, p = 0.032) and VO(2peak) (t = 2.4, p = 0.037) were better predictors for phosphorylated AS160 than TNF-alpha or IL-6 (p>0.05). Conclusions: Elevated inflammatory markers and increased metabolic risk factors may inhibit insulin-signaling protein phosphorylation in middle-aged women, thereby increasing insulin resistance under basal conditions. Furthermore, higher HDL and fitness levels are associated with an increased AS160 phosphorylation, which may in turn reduce insulin resistance.

Reactions of the hydroperoxide anion with dimethyl methylphosphonate in an ion trap mass spectrometer : evidence for a gas phase alpha-effect

The gas phase degradation reactions of the chemical warfare agent (CWA) simulant, dimethyl methylphosphonate (DMMP), with the hydroperoxide anion (HOO(-)) were investigated using a modified quadrupole ion trap mass spectrometer. The HOO(-) anion reacts readily with neutral DMMP forming two significant product ions at m/z 109 and m/z 123. The major reaction pathways correspond to (i) the nucleophilic substitution at carbon to form \[CH(3)P(O)(OCH(3))O](-) (m/z 109) in a highly exothermic process and (ii) exothermic proton transfer. The branching ratios of the two reaction pathways, 89% and 11% respectively, indicate that the former reaction is significantly faster than the latter. This is in contrast to the trend for the methoxide anion with DMMP, where proton transfer dominates. The difference in the observed reactivities of the HOO(-) and CH(3)O(-) anions can be considered as evidence for an a-effect in the gas phase and is supported by electronic structure calculations at the B3LYP/aug-cc-pVTZ//B3LYP/6-31+G(d) level of theory that indicate the S(N)2(carbon) process has an activation energy 7.8 kJ mol(-1) lower for HOO(-) as compared to CH(3)O(-). A similar alpha-effect was calculated for nucleophilic addition-elimination at phosphorus, but this process an important step in the perhydrolysis degradation of CWAs in solution - was not observed to occur with DMMP in the gas phase. A theoretical investigation revealed that all processes are energetically accessible with negative activation energies. However, comparison of the relative Arrhenius pre-exponential factors indicate that substitution at phosphorus is not kinetically competitive with respect to the S(N)2(carbon) and deprotonation processes.