Automatic labelling of topic models learned from Twitter by summarisation

Latent topics derived by topic models such as Latent Dirichlet Allocation (LDA) are the result of hidden thematic structures which provide further insights into the data. The automatic labelling of such topics derived from social media poses however new challenges since topics may characterise novel events happening in the real world. Existing automatic topic labelling approaches which depend on external knowledge sources become less applicable here since relevant articles/concepts of the extracted topics may not exist in external sources. In this paper we propose to address the problem of automatic labelling of latent topics learned from Twitter as a summarisation problem. We introduce a framework which apply summarisation algorithms to generate topic labels. These algorithms are independent of external sources and only rely on the identification of dominant terms in documents related to the latent topic. We compare the efficiency of existing state of the art summarisation algorithms. Our results suggest that summarisation algorithms generate better topic labels which capture event-related context compared to the top-n terms returned by LDA. © 2014 Association for Computational Linguistics.

Techniques for automated taxonomy building: towards ontologies for knowledge management

Ontologies have become widely accepted as the main method for representing knowledge in Knowledge Management (KM) applica-tions. Given the continuous and rapid change and dynamic nature of knowledge in all fields, automated methods for construct-ing ontologies are of great importance. All ontologies or taxonomies currently in use have been hand built and require consider-able manpower to keep up to date. Taxono-mies are less logically rigorous than ontolo-gies, and in this paper we consider the re-quirements for a system which automatically constructed taxonomies. There are a number of potentially useful methods for construct-ing hierarchically organised concepts from a collection of texts and there are a number of automatic methods which permit one to as-sociate one word with another. The impor-tant issue for the successful development of this research area is to identify techniques for labelling the relation between two candi-date terms, if one exists. We consider a number of possible approaches and argue that the majority are unsuitable for our re-quirements.

Ontologies, taxonomies, thesauri:learning from texts

The use of ontologies as representations of knowledge is widespread but their construction, until recently, has been entirely manual. We argue in this paper for the use of text corpora and automated natural language processing methods for the construction of ontologies. We delineate the challenges and present criteria for the selection of appropriate methods. We distinguish three ma jor steps in ontology building: associating terms, constructing hierarchies and labelling relations. A number of methods are presented for these purposes but we conclude that the issue of data-sparsity still is a ma jor challenge. We argue for the use of resources external tot he domain specific corpus.

Biomarkers

Introduction – Why do we need ‘biomarkers? Biomarkers of protein oxidation Introduction Major issues/questions Protein carbonyl biomarkers Biochemistry Methods of measurement Storage, stability and limitations in use Protein thiol biomarkers Biochemistry Methods of measurement Storage, stability and limitations on use Aliphatic amino acid biomarkers Biochemistry Methods of measurement Storage, stability and limitations on use Oxidised Tryptophan Biomarkers Biochemistry Method of measurement Storage, stability and limitations on use Oxidised tyrosine biomarkers Biochemistry Methods of measurement Storage, stability and limitations on use Formation of neoepitopes on oxidised proteins Validation of assays for protein oxidation biomarkers Relationship of protein oxidation to disease Modulation of protein oxidation biomarkers by antioxidants Future perspectives Introduction to lipid peroxidation biomarkers Introduction: biochemistry of lipid peroxidation Malondialdehyde Methods of measurement Storage, stability and limitations on use Conjugated dienes Method of measurement Storage, stability and limitations of use LDL lag phase Method of measurement Storage, stability and limitations of use Hydrocarbon gases Biochemistry Method of measurement Storage, stability and limitations on use Lipofuscin Biochemistry Method of measurement Storage, stability and limitation on use Lipid peroxides Biochemistry Method of measurement Storage, stability and limitations on use Isoprostanes Biochemistry Method of measurement Storage, stability and limitations on use Possible new biomarkers of lipid oxidation Relationship of lipid peroxidation to disease Modulation of lipid peroxidation biomarkers by antioxidants Functional consequences of lipid peroxidation Contribution of dietary intake to lipid peroxidation products Biomarkers of DNA oxidation Introduction Confounding factors Units and terminology Nuclear and mitochondrial DNA damage Lymphocytes as surrogate tissues Measurement of DNA damage with the comet assay Practical details Storage, stability, and limitations of the assay Measurement of DNA base oxidation by HPLC Practical details Storage, stability and limitations of the method Measurement of DNA base oxidation by GC–MS Biochemistry of 8-oxoguanine, adenine and fapy derivatives Methods of measurement Storage, stability and limitations of the method Analysis of guanine oxidation products in urine Method of measurement Limitations and criticisms Immunochemical methods Methods of measurement Storage, stability, and limitations of the assay 32P post-labelling Method of measurement Limitations and criticisms Validation of assays for DNA oxidation Oxo-dGuo in lymphocyte DNA Urinary measurements DNA–aldehyde adducts Biochemistry Method of measurement Products of reactive nitrogen species Endpoints arising from oxidative DNA damage Mutations Chromosome aberrations Micronuclei Site-specific DNA damage Relationship of DNA oxidation to disease Modulation of DNA oxidation biomarkers by antioxidants Direct and indirect effects of oxidative stress: measures of total oxidant/antioxidant levels Visualisation of cellular oxidants Biochemistry: histochemical detection of ROS Method of measurement Limitations, storage and stability Measurement of hydrogen peroxide Biochemistry Methods of measurement Storage, stability and limitations on use Measurement of the ratio of antioxidant/oxidised antioxidant Biochemistry Method of measurement Storage, stability and limitations on use Total antioxidant capacity Biochemistry Terminology Methods of measurement Storage, stability and limitations on use Validation of assays for direct oxidant and antioxidant biomarkers Relationship of oxidant/antioxidant measurement to disease Modulation of oxidant/antioxidant biomarkers by dietary antioxidants Induction of genes in response to oxidative stress Background Measurement of antioxidant responsive genes and proteins Effects of antioxidant intake on the activity of antioxidant enzymes

Eukaryotic expression:Developments for structural proteomics

The production of sufficient quantities of protein is an essential prelude to a structure determination, but for many viral and human proteins this cannot be achieved using prokaryotic expression systems. Groups in the Structural Proteomics In Europe (SPINE) consortium have developed and implemented high-throughput (HTP) methodologies for cloning, expression screening and protein production in eukaryotic systems. Studies focused on three systems: yeast (Pichia pastoris and Saccharomyces cerevisiae), baculovirus-infected insect cells and transient expression in mammalian cells. Suitable vectors for HTP cloning are described and results from their use in expression screening and protein-production pipelines are reported. Strategies for co-expression, selenomethionine labelling (in all three eukaryotic systems) and control of glycosylation (for secreted proteins in mammalian cells) are assessed. © International Union of Crystallography, 2006.

Towards sea surface pollution detection from visible band images

This paper presents a novel approach to water pollution detection from remotely sensed low-platform mounted visible band camera images. We examine the feasibility of unsupervised segmentation for slick (oily spills on the water surface) region labelling. Adaptive and non adaptive filtering is combined with density modeling of the obtained textural features. A particular effort is concentrated on the textural feature extraction from raw intensity images using filter banks and adaptive feature extraction from the obtained output coefficients. Segmentation in the extracted feature space is achieved using Gaussian mixture models (GMM).

Functional presynaptic HCN channels in the rat globus pallidus

Hyperpolarization-activated, cyclic nucleotide-gated cation (HCN) channels are expressed postsynaptically in the rodent globus pallidus (GP), where they play several important roles in controlling GP neuronal activity. To further elucidate the role of HCN channels in the GP, immunocytochemical and electrophysiological approaches were used to test the hypothesis that HCN channels are also expressed presynaptically on the local axon collaterals of GP neurons. At the electron microscopic level, immunoperoxidase labelling for HCN1 and HCN2 was localized in GP somata and dendritic processes, myelinated and unmyelinated axons, and axon terminals. One population of labelled terminals formed symmetric synapses with somata and proximal dendrites and were immunoreactive for parvalbumin, consistent with the axon collaterals of GABAergic GP projection neurons. In addition, labelling for HCN2 and, to a lesser degree, HCN1 was observed in axon terminals that formed asymmetric synapses and were immunoreactive for the vesicular glutamate transporter 2. Immunogold labelling demonstrated that HCN1 and HCN2 were located predominantly at extrasynaptic sites along the plasma membrane of both types of terminal. To determine the function of presynaptic HCN channels in the GP, we performed whole-cell recordings from GP neurons in vitro. Bath application of the HCN channel blocker ZD7288 resulted in an increase in the frequency of mIPSCs but had no effect on their amplitude, implying that HCN channels tonically regulate the release of GABA. Their presence, and predicted role in modulating transmitter release, represents a hitherto unidentified mechanism whereby HCN channels influence the activity of GP neurons. © The Authors (2007).

Development of a quality control material for the measurement of 8-oxo-7,8-dihydro-2'-deoxyguanosine, an in vivo marker of oxidative stress, and comparison of results from different laboratories

The measurement of 8-oxo-7,8-dihydro-2'-deoxyguanosine is an increasingly popular marker of in vivo oxidative damage to DNA. A random-sequence 21-mer oligonucleotide 5'-TCA GXC GTA CGT GAT CTC AGT-3' in which X was 8-oxo-guanine (8-oxo-G) was purified and accurate determination of the oxidised base was confirmed by a 32P-end labelling strategy. The lyophilised material was analysed for its absolute content of 8-oxo-dG by several major laboratories in Europe and one in Japan. Most laboratories using HPLC-ECD underestimated, while GC-MS-SIM overestimated the level of the lesion. HPLC-ECD measured the target value with greatest accuracy. The results also suggest that none of the procedures can accurately quantitate levels of 1 in 10(6) 8-oxo-(d)G in DNA.

Facial emotion recognition and alexithymia in adults with somatoform disorders

The primary aim of this study was to investigate facial emotion recognition (FER) in patients with somatoform disorders (SFD). Also of interest was the extent to which concurrent alexithymia contributed to any changes in emotion recognition accuracy. Twenty patients with SFD and 20 healthy, age, sex and education matched, controls were assessed with the Facially Expressed Emotion Labelling Test of FER and the 26-item Toronto Alexithymia Scale. Patients withSFD exhibited elevated alexithymia symptoms relative to healthy controls.Patients with SFD also recognized significantly fewer emotional expressions than did the healthy controls. However, the group difference in emotion recognition accuracy became nonsignificant once the influence of alexithymia was controlled for statistically. This suggests that the deficit in FER observed in the patients with SFD was most likely a consequence of concurrent alexithymia. It should be noted that neither depression nor anxiety was significantly related to emotion recognition accuracy, suggesting that these variables did not contribute the emotion recognition deficit. Impaired FER observed in the patients with SFD could plausibly have a negative influence on these individuals’ social functioning.

Modulation of tissue transglutaminase in tubular epithelial cells alters extracellular matrix levels: a potential mechanism of tissue scarring

The up-regulation and trafficking of tissue transglutaminase (TG2) by tubular epithelial cells (TEC) has been implicated in the development of kidney scarring. TG2 catalyses the crosslinking of proteins via the formation of highly stable e(?-glutamyl) lysine bonds. We have proposed that TG2 may contribute to kidney scarring by accelerating extracellular matrix (ECM) deposition and by stabilising the ECM against proteolytic decay. To investigate this, we have studied ECM metabolism in Opossum kidney (OK) TEC induced to over-express TG2 by stable transfection and in tubular cells isolated from TG2 knockout mice. Increasing the expression of TG2 led to increased extracellular TG2 activity (p < 0.05), elevated e(?-glutamyl) lysine crosslinking in the ECM and higher levels of ECM collagen per cell by 3H-proline labelling. Immunofluorescence demonstrated that this was attributable to increased collagen III and IV levels. Higher TG2 levels were associated with an accelerated collagen deposition rate and a reduced ECM breakdown by matrix metalloproteinases (MMPs). In contrast, a lack of TG2 was associated with reduced e(?-glutamyl) lysine crosslinking in the ECM, causing reduced ECM collagen levels and lower ECM per cell. We report that TG2 contributes to ECM accumulation primarily by accelerating collagen deposition, but also by altering the susceptibility of the tubular ECM to decay. These findings support a role for TG2 in the expansion of the ECM associated with kidney scarring.

Spatial aspects of MRSA epidemiology:A case study using stochastic simulation, kernel estimation and SaTScan

The identification of disease clusters in space or space-time is of vital importance for public health policy and action. In the case of methicillin-resistant Staphylococcus aureus (MRSA), it is particularly important to distinguish between community and health care-associated infections, and to identify reservoirs of infection. 832 cases of MRSA in the West Midlands (UK) were tested for clustering and evidence of community transmission, after being geo-located to the centroids of UK unit postcodes (postal areas roughly equivalent to Zip+4 zip code areas). An age-stratified analysis was also carried out at the coarser spatial resolution of UK Census Output Areas. Stochastic simulation and kernel density estimation were combined to identify significant local clusters of MRSA (p<0.025), which were supported by SaTScan spatial and spatio-temporal scan. In order to investigate local sampling effort, a spatial 'random labelling' approach was used, with MRSA as cases and MSSA (methicillin-sensitive S. aureus) as controls. Heavy sampling in general was a response to MRSA outbreaks, which in turn appeared to be associated with medical care environments. The significance of clusters identified by kernel estimation was independently supported by information on the locations and client groups of nursing homes, and by preliminary molecular typing of isolates. In the absence of occupational/ lifestyle data on patients, the assumption was made that an individual's location and consequent risk is adequately represented by their residential postcode. The problems of this assumption are discussed, with recommendations for future data collection.

Protein oxidation:role in signalling and detection by mass spectrometry

Proteins can undergo a wide variety of oxidative post-translational modifications (oxPTM); while reversible modifications are thought to be relevant in physiological processes, non-reversible oxPTM may contribute to pathological situations and disease. The oxidant is also important in determining the type of oxPTM, such as oxidation, chlorination or nitration. The best characterized oxPTMs involved in signalling modulation are partial oxidations of cysteine to disulfide, glutathionylated or sulfenic acid forms that can be reversed by thiol reductants. Proline hydroxylation in HIF signalling is also quite well characterized, and there is increasing evidence that specific oxidations of methionine and tyrosine may have some biological roles. For some proteins regulated by cysteine oxidation, the residues and molecular mechanism involved have been extensively studied and are well understood, such as the protein tyrosine phosphatase PTP1B and MAP3 kinase ASK1, as well as transcription factor complex Keap1-Nrf2. The advances in understanding of the role oxPTMs in signalling have been facilitated by advances in analytical technology, in particular tandem mass spectrometry techniques. Combinations of peptide sequencing by collisionally induced dissociation and precursor ion scanning or neutral loss to select for specific oxPTMs have proved very useful for identifying oxidatively modified proteins and mapping the sites of oxidation. The development of specific labelling and enrichment procedures for S-nitrosylation or disulfide formation has proved invaluable, and there is ongoing work to establish analogous methods for detection of nitrotyrosine and other modifications.

Investigation and development of techniques for the characterisation of the synthetic/biological interface

The purpose of this study is to increase our knowledge of the nature of the surface properties of polymeric materials and improve our understanding of how these factors influence the deposition of proteins to form a reactive biological/synthetic interface. A number of surface analytical techniques were identified as being of potential benefit to this investigation and included in a multidisciplinary research program. Cell adhesion in culture was the primary biological sensor of surface properties, and it showed that the cell response to different materials can be modified by adhesion promoting protein layers: cell adhesion is a protein-mediated event. A range of surface rugosity can be produced on polystyrene, and the results presented here show that surface rugosity does not play a major role in determining a material's cell adhesiveness. Contact angle measurements showed that surface energy (specifically the polar fraction) is important in promoting cell spreading on surfaces. The immunogold labelling technique indicated that there were small, but noticeable differences, between the distribution of proteins on a range of surfaces. This study has shown that surface analysis techniques have different sensitivities in terms of detection limits and depth probed, and these are important in determining the usefulness of the information obtained. The techniques provide information on differing aspects of the biological/synthetic interface, and the consequence of this is that a range of techniques is needed in any full study of such a complex field as the biomaterials area.

Cationic liposomes as adjuvants for subunit protein vaccines:their role in the depot effect and antigen processing by APCs

Introduction: The requirement of adjuvants in subunit protein vaccination is well known yet their mechanisms of action remain elusive. Of the numerous mechanisms suggested, cationic liposomes appear to fulfil at least three: the antigen depot effect, the delivery of antigen to antigen presenting cells (APCs) and finally the danger signal. We have investigated the role of antigen depot effect with the use of dual radiolabelling whereby adjuvant and antigen presence in tissues can be quantified. In our studies a range of cationic liposomes and different antigens were studied to determine the importance of physical properties such as liposome surface charge, antigen association and inherent lipid immunogenicity. More recently we have investigated the role of liposome size with the cationic liposome formulation DDA:TDB, composed of the cationic lipid dimethyldioctadecylammonium (DDA) and the synthetic mycobacterial glycolipid trehalose 6,6’-dibehenate (TDB). Vesicle size is a frequently investigated parameter which is known to result in different routes of endocytosis. It has been postulated that targeting different routes leads to different intracellular signaling pathway activation and it is certainly true that numerous studies have shown vesicle size to have an effect on the resulting immune responses (e.g. Th1 vs. Th2). Aim: To determine the effect of cationic liposome size on the biodistribution of adjuvant and antigen, the ensuing humoral and cell-mediated immune responses and the uptake and activation of antigen by APCs including macrophages and dendritic cells. Methods: DDA:TDB liposomes were made to three different sizes (~ 0.2, 0.5 and 2 µm) followed by the addition of tuberculosis antigen Ag85B-ESAT-6 therefore resulting in surface adsorption. Liposome formulations were injected into Balb/c or C57Bl/6 mice via the intramuscular route. The biodistribution of the liposome formulations was followed using dual radiolabelling. Tissues including muscle from the site of injection and local draining lymph nodes were removed and liposome and antigen presence quantified. Mice were also immunized with the different vaccine formulations and cytokine production (from Ag85B-ESAT-6 restimulated splenocytes) and antibody presence in blood assayed. Furthermore, splenocyte proliferation after restimulating with Ag85B-ESAT-6 was measured. Finally, APCs were compared for their ability to endocytose vaccine formulations and the effect this had on the maturation status of the cell populations was compared. Flow cytometry and fluorescence labelling was used to investigate maturation marker up-regulation and efficacy of phagocytosis. Results: Our results show that for an efficient Ag85B-ESAT-6 antigen depot at the injection site, liposomes composed of DDA and TDB are required. There is no significant change in the presence of liposome or antigen at 6hrs or 24hrs p.i, nor does liposome size have an effect. Approximately 0.05% of the injected liposome dose is detected in the local draining lymph node 24hrs p.i however protein presence is low (<0.005% dose). Preliminary in vitro data shows liposome and antigen endocytosis by macrophages; further studies on this will be presented in addition to the results of the immunisation study.

Studies on the generalisation of Gaussian processes and Bayesian neural networks

The assessment of the reliability of systems which learn from data is a key issue to investigate thoroughly before the actual application of information processing techniques to real-world problems. Over the recent years Gaussian processes and Bayesian neural networks have come to the fore and in this thesis their generalisation capabilities are analysed from theoretical and empirical perspectives. Upper and lower bounds on the learning curve of Gaussian processes are investigated in order to estimate the amount of data required to guarantee a certain level of generalisation performance. In this thesis we analyse the effects on the bounds and the learning curve induced by the smoothness of stochastic processes described by four different covariance functions. We also explain the early, linearly-decreasing behaviour of the curves and we investigate the asymptotic behaviour of the upper bounds. The effect of the noise and the characteristic lengthscale of the stochastic process on the tightness of the bounds are also discussed. The analysis is supported by several numerical simulations. The generalisation error of a Gaussian process is affected by the dimension of the input vector and may be decreased by input-variable reduction techniques. In conventional approaches to Gaussian process regression, the positive definite matrix estimating the distance between input points is often taken diagonal. In this thesis we show that a general distance matrix is able to estimate the effective dimensionality of the regression problem as well as to discover the linear transformation from the manifest variables to the hidden-feature space, with a significant reduction of the input dimension. Numerical simulations confirm the significant superiority of the general distance matrix with respect to the diagonal one.In the thesis we also present an empirical investigation of the generalisation errors of neural networks trained by two Bayesian algorithms, the Markov Chain Monte Carlo method and the evidence framework; the neural networks have been trained on the task of labelling segmented outdoor images.