Learning Methods and Algorithms for Semantic Text Classification across Multiple Domains

Information is nowadays a key resource: machine learning and data mining techniques have been developed to extract high-level information from great amounts of data. As most data comes in form of unstructured text in natural languages, research on text mining is currently very active and dealing with practical problems. Among these, text categorization deals with the automatic organization of large quantities of documents in priorly defined taxonomies of topic categories, possibly arranged in large hierarchies. In commonly proposed machine learning approaches, classifiers are automatically trained from pre-labeled documents: they can perform very accurate classification, but often require a consistent training set and notable computational effort. Methods for cross-domain text categorization have been proposed, allowing to leverage a set of labeled documents of one domain to classify those of another one. Most methods use advanced statistical techniques, usually involving tuning of parameters. A first contribution presented here is a method based on nearest centroid classification, where profiles of categories are generated from the known domain and then iteratively adapted to the unknown one. Despite being conceptually simple and having easily tuned parameters, this method achieves state-of-the-art accuracy in most benchmark datasets with fast running times. A second, deeper contribution involves the design of a domain-independent model to distinguish the degree and type of relatedness between arbitrary documents and topics, inferred from the different types of semantic relationships between respective representative words, identified by specific search algorithms. The application of this model is tested on both flat and hierarchical text categorization, where it potentially allows the efficient addition of new categories during classification. Results show that classification accuracy still requires improvements, but models generated from one domain are shown to be effectively able to be reused in a different one.

Multiple domains of repressor activator protein 1 contribute to facilitated binding of glycolysis regulatory protein 1

The function of repressor activator protein 1 (Rap1p) at glycolytic enzyme gene upstream activating sequence (UAS) elements in Saccharomyces cerevisiae is to facilitate binding of glycolysis regulatory protein 1 (Gcr1p) at adjacent sites. Rap1p has a modular domain structure. In its amino terminus there is an asymmetric DNA-bending domain, which is distinct from its DNA-binding domain, which resides in the middle of the protein. In the carboxyl terminus of Rap1p lie its silencing and putative activation domains. We carried out a molecular dissection of Rap1p to identify domains contributing to its ability to facilitate binding of Gcr1p. We prepared full-length and three truncated versions of Rap1p and tested their ability to facilitate binding of Gcr1p by gel shift assay. The ability to detect ternary complexes containing Rap1p⋅DNA⋅Gcr1p depended on the presence of binding sites for both proteins in the probe DNA. The DNA-binding domain of Rap1p, although competent to bind DNA, was unable to facilitate binding of Gcr1p. Full-length Rap1p and the amino- and carboxyl-truncated versions of Rap1p were each able to facilitate binding of Gcr1p at an appropriately spaced binding site. Under these conditions, Gcr1p displayed an approximately 4-fold greater affinity for Rap1p-bound DNA than for otherwise identical free DNA. When spacing between Rap1p- and Gcr1p-binding sites was altered by insertion of five nucleotides, the ability to form ternary Rap1p⋅DNA⋅Gcr1p complexes was inhibited by all but the DNA-binding domain of Rap1p itself; however, the ability of each individual protein to bind the DNA probe was unaffected.

Multiple Domains of Fission Yeast Cdc19p (MCM2) Are Required for Its Association with the Core MCM Complex

The members of the MCM protein family are essential eukaryotic DNA replication factors that form a six-member protein complex. In this study, we use antibodies to four MCM proteins to investigate the structure of and requirements for the formation of fission yeast MCM complexes in vivo, with particular regard to Cdc19p (MCM2). Gel filtration analysis shows that the MCM protein complexes are unstable and can be broken down to subcomplexes. Using coimmunoprecipitation, we find that Mis5p (MCM6) and Cdc21p (MCM4) are tightly associated with one another in a core complex with which Cdc19p loosely associates. Assembly of Cdc19p with the core depends upon Cdc21p. Interestingly, there is no obvious change in Cdc19p-containing MCM complexes through the cell cycle. Using a panel of Cdc19p mutants, we find that multiple domains of Cdc19p are required for MCM binding. These studies indicate that MCM complexes in fission yeast have distinct substructures, which may be relevant for function.

Multiple protein domains determine the cell type-specific nuclear distribution of the catalytic subunit required for apolipoprotein B mRNA editing

Apolipoprotein B (apoB) mRNA editing catalyzed by apoB mRNA editing catalytic subunit 1 (APOBEC-1) has been proposed to be a nuclear process. To test this hypothesis, the subcellular distribution of hemagglutinin-(HA) tagged APOBEC-1 expressed in transiently transfected hepatoma cells was determined by indirect immunofluorescence microscopy. HA-APOBEC-1 was detected in both the nucleus and cytoplasm of rat and human hepatoma cells. Mutagenesis of APOBEC-1 demonstrated that the N-terminal 56 amino acids (1–56) were necessary for the nuclear distribution of APOBEC-1, but this region did not contain a functional nuclear localization signal (NLS). However, we identified a 24-amino acid domain in the C terminus of APOBEC-1 with characteristics of a cytoplasmic retention signal (CRS) or a nuclear export signal (NES). These data suggest, therefore, that the nuclear import of APOBEC-1 may not be mediated by a positive NLS; rather, it may be achieved by overcoming the effect of a CRS/NES. We also demonstrated that the nuclear distribution of APOBEC-1 occurred only in cell lines that were capable of editing apoB RNA. We propose that the cellular distribution of APOBEC-1 is determined by multiple domains within this protein, and a nuclear localization of the enzyme may be regulated by cell type-specific factors that render these cells uniquely editing competent.

The contribution of different androgen receptor domains to receptor dimerization and signaling

The androgen receptor (AR) is a ligand-activated transcription factor of the nuclear receptor superfamily that plays a critical role in male physiology and pathology. Activated by binding of the native androgens testosterone and 5-dihydrotestosterone, the AR regulates transcription of genes involved in the development and maintenance of male phenotype and male reproductive function as well as other tissues such as bone and muscle. Deregulation of AR signaling can cause a diverse range of clinical conditions, including the X-linked androgen insensitivity syndrome, a form of motor neuron disease known as Kennedy’s disease, and male infertility. In addition, there is now compelling evidence that the AR is involved in all stages of prostate tumorigenesis including initiation, progression, and treatment resistance. To better understand the role of AR signaling in the pathogenesis of these conditions, it is important to have a comprehensive understanding of the key determinants of AR structure and function. Binding of androgens to the AR induces receptor dimerization, facilitating DNA binding and the recruitment of cofactors and transcriptional machinery to regulate expression of target genes. Various models of dimerization have been described for the AR, the most well characterized interaction being DNA-binding domain- mediated dimerization, which is essential for the AR to bind DNA and regulate transcription. Additional AR interactions with potential to contribute to receptor dimerization include the intermolecular interaction between the AR amino terminal domain and ligand-binding domain known as the N-terminal/C-terminal interaction, and ligand-binding domain dimerization. In this review, we discuss each form of dimerization utilized by the AR to achieve transcriptional competence and highlight that dimerization through multiple domains is necessary for optimal AR signaling.

Signal peptide and helical bundle domains of virulent canine distemper virus fusion protein restrict fusogenicity

Persistence in canine distemper virus (CDV) infection is correlated with very limited cell-cell fusion and lack of cytolysis induced by the neurovirulent A75/17-CDV compared to that of the cytolytic Onderstepoort vaccine strain. We have previously shown that this difference was at least in part due to the amino acid sequence of the fusion (F) protein (P. Plattet, J. P. Rivals, B. Zuber, J. M. Brunner, A. Zurbriggen, and R. Wittek, Virology 337:312-326, 2005). Here, we investigated the molecular mechanisms of the neurovirulent CDV F protein underlying limited membrane fusion activity. By exchanging the signal peptide between both F CDV strains or replacing it with an exogenous signal peptide, we demonstrated that this domain controlled intracellular and consequently cell surface protein expression, thus indirectly modulating fusogenicity. In addition, by serially passaging a poorly fusogenic virus and selecting a syncytium-forming variant, we identified the mutation L372W as being responsible for this change of phenotype. Intriguingly, residue L372 potentially is located in the helical bundle domain of the F(1) subunit. We showed that this mutation drastically increased fusion activity of F proteins of both CDV strains in a signal peptide-independent manner. Due to its unique structure even among morbilliviruses, our findings with respect to the signal peptide are likely to be specifically relevant to CDV, whereas the results related to the helical bundle add new insights to our growing understanding of this class of F proteins. We conclude that different mechanisms involving multiple domains of the neurovirulent A75/17-CDV F protein act in concert to limit fusion activity, preventing lysis of infected cells, which ultimately may favor viral persistence.

The cDNA cloning of murine HIP /L29 and the functional analysis of HIP/L29 domains

Human heparin/heparan sulfate interacting protein/L29 (HIP/L29) is a heparin/heparan sulfate (Hp/HS) binding protein found in many adult human tissues. Potential functions of this protein are promotion of embryo adhesion, modulation of blood coagulation, and control of cell growth. While these activities are diverse, the ability of human HIP/L29 to interact with Hp/HS at the cell surface may be a unifying mechanism of action since Hp/HS influences all of these processes. A murine ortholog has been identified that has 78.8% homology over the entire sequence and identity over the N-terminal 64 amino acids when compared to human HIP/L29. Northern, Western, and immunohistochemical analysis shows that murine HIP/L29 mRNA and protein are expressed in a tissue specific manner. Murine HIP/L29 is enriched in the membrane fraction of NmuMG cells where it is eluted with high salt, suggesting that it is a peripheral membrane protein. The ability of murine HIP/L29 to bind Hp is verified by studies using native and recombinant forms of murine HIP/L29. A synthetic peptide (HIP peptide-2) derived from the identical N-terminal region of HIP/L29 proteins was tested for the ability to bind Hp and support cell adhesion. This peptide was chosen because it conforms to a proposed consensus sequence for Hp/HS binding peptides. HIP peptide-2 binds Hp in a dose-dependent, saturable, and selective manner and supports Hp-dependent cell adhesion. However, a scrambled form of this peptide displayed similar activities indicating a lack of peptide sequence specificity required for activity. Lastly, an unbiased approach was used to identify sequences within human and mouse HIP/L29 proteins necessary for Hp/HS binding. A panel of recombinant proteins was made that collectively are deficient in every human HIP/L29 domain. The activities of these deletion mutants and recombinant murine HIP/L29 were compared to the activity of recombinant human HIP/L29 in a number of assays designed to look at differences in the ability to bind Hp/HS. These studies suggest that each domain within human HIP/L29 is important for binding to Hp/HS and divergences in the C-terminus of human and mouse HIP/L29 account for a decrease in murine HIP/L29 affinity for Hp/HS. It is apparent that multiple domains within human and mouse HIP/L29 contribute to the function of Hp/HS binding. The interaction of multiple HIP/L29 domains with Hp/HS will influence the biological activity of HIP/L29 proteins. ^

The Multidimensional Loss Scale : validating a cross cultural instrument for measuring loss in Burmese refugees

The Multidimensional Loss Scale: Initial Development and Psychometric Evaluation The Multidimensional Loss Scale (MLS) represents the first instrument designed specifically to measure loss in refugee populations. Researchers developed initial items of the Multidimensional Loss Scale to assess Experience of Loss Events and Loss Distress in a culturally sensitive manner across multiple domains (social, material, intra-personal and cultural). A sample of 70 recently settled Burmese adult refugees completed a battery of questionnaires, including new scale items. Analyses explored the scale’s factor structure, internal consistency, convergent validity and divergent validity. Principal Axis Factoring supported a five-factor model: Loss of Symbolic Self, Loss of Interdependence, Loss of Home, Interpersonal Loss, and Loss of Intrapersonal Integrity. Chronbach’s Alphas indicated satisfactory internal consistency for Experience of Loss Events (.85) and Loss Distress (.92). Convergent and divergent validity of Loss Distress were supported by moderate correlations with interpersonal grief and trauma symptoms and weak correlations with depression and anxiety. The new scale was well received by people from refugee backgrounds and shows promise for application in future research and practice

Natural recovery from cannabis use in people with psychosis: A qualitative study

- Objective There is rapidly growing evidence of natural recovery from cannabis use in people with psychosis, but little is known about how it occurs. This qualitative study explores what factors influence the decision to cease cannabis use, maintain cessation, and prevent relapse. - Methods Ten people with early psychosis and lifetime cannabis misuse, who had been abstinent for at least a month, were recruited from public adult mental health services. These six men and four women participated in a semi-structured qualitative interview assessing reasons for addressing cannabis use, effective change strategies, lapse contexts, and methods used to regain control. Interpretative phenomenological analysis was used to identify themes in their responses. - Results Participants had a mean age of 23 years (SD = 3.7), started using cannabis at age 13.7 (SD = 1.6), began daily use at 17 (SD = 3.1), and had abstained from cannabis for 7.9 months (SD = 5.4). Awareness of the negative impact of substance use across multiple domains and the presence of social support for cannabis cessation were seen as vital to sustained success, as was utilization of a combination of coping strategies. The ability to address pressure from substance-using peers was commonly mentioned. - Conclusions Maximally effective treatment may need to focus on eliciting a range of benefits of cessation and control strategies and on maximizing both support for change and resistance to peer pressure. Further research might focus on comparing perceived effective strategies between individuals who obtain sustained cessation versus those who relapse.

The evolution of information while building cross-domain models of a design: A video experiment

Models capturing the connectivity between different domains of a design, e.g. between components and functions, can provide a tool for tracing and analysing aspects of that design. In this paper, video experiments are used to explore the role of cross-domain modelling in building up information about a design. The experiments highlight that cross-domain modelling can be a useful tool to create and structure design information. Findings suggest that consideration of multiple domains encourages discussion during modelling, helps identify design aspects that might otherwise be overlooked, and can help promote consideration of alternative design options. Copyright © 2002-2012 The Design Society. All rights reserved.

Cognitive Adaptation to the Experience of Social and Cultural Diversity

Diversity is a defining characteristic of modern society, yet there remains considerable debate over the benefits that it brings. The authors argue that positive psychological and behavioral outcomes will be observed only when social and cultural diversity is experienced in a way that challenges stereotypical expectations and that when this precondition is met, the experience has cognitive consequences that resonate across multiple domains. A model, rooted in social categorization theory and research, outlines the preconditions and processes through which people cognitively adapt to the experience of social and cultural diversity and the resulting cross-domain benefits that this brings. Evidence is drawn from a range of literatures to support this model, including work on biculturalism, minority influence, cognitive development, stereotype threat, work group productivity, creativity, and political ideology. The authors bring together a range of differing diversity experiences and explicitly draw parallels between programs of research that have focused on both perceiving others who are multicultural and being multicultural oneself. The findings from this integrative review suggest that experiencing diversity that challenges expectations may not only encourage greater tolerance but also have benefits beyond intergroup relations to varied aspects of psychological functioning.

Episodic memory and learning in patients with chronic schizophrenia

This study explored the pattern of memory functioning in 58 patients with chronic schizophrenia and compared their performance with 53 normal controls. Multiple domains of memory were assessed, including verbal and nonverbal memory span, verbal and non-verbal paired associate learning, verbal and visual long-term memory, spatial and non-spatial conditional associative learning, recognition memory and memory for temporal order. Consistent with previous studies, substantial deficits in long-term memory were observed, with relative preservation of memory span. Memory for temporal order and recognition memory was intact, although significant deficits were observed on the conditional associative learning tasks. There was no evidence of lateralized memory impairment. In these respects, the pattern of memory impairment in schizophrenia is more similar in nature to that found in patients with memory dysfunction following mesiotemporal lobe lesions, rather than that associated with focal frontal lobe damage. (C) 1999 Elsevier Science B.V. All rights reserved.

A shared care model pilot for palliative home care in a rural area:Impact on symptoms, distress, and place of death

Context: Shared care models integrating family physician services with interdisciplinary palliative care specialist teams are critical to improve access to quality palliative home care and address multiple domains of end-of-life issues and needs. Objectives: To examine the impact of a shared care pilot program on the primary outcomes of symptom severity and emotional distress (patient and family separately) over time and, secondarily, the concordance between patient preferences and place of death. Methods: An inception cohort of patients (n = 95) with advanced, progressive disease, expected to die within six months, were recruited from three rural family physician group practices (21 physicians) and followed prospectively until death or pilot end. Serial measurement of symptoms, emotional distress (patient and family), and preferences for place of death was performed, with analysis of changes in distress outcomes assessed using t-tests and general linear models. Results: Symptoms trended toward improvement, with a significant reduction in anxiety from baseline to 14 days noted. Symptom and emotional distress were maintained below high severity (7-10), and a high rate of home death compared with population norms was observed. Conclusion: Future controlled studies are needed to examine outcomes for shared care models with comparison groups. Shared care models build on family physician capacity and as such are promising in the development of palliative home care programs to improve access to quality palliative home care and foster health system integration. © 2011 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved.

Association Study of 167 Candidate Genes for Schizophrenia Selected by a Multi-Domain Evidence-Based Prioritization Algorithm and Neurodevelopmental Hypothesis

Integrating evidence from multiple domains is useful in prioritizing disease candidate genes for subsequent testing. We ranked all known human genes (n = 3819) under linkage peaks in the Irish Study of High-Density Schizophrenia Families using three different evidence domains: 1) a meta-analysis of microarray gene expression results using the Stanley Brain collection, 2) a schizophrenia protein-protein interaction network, and 3) a systematic literature search. Each gene was assigned a domain-specific p-value and ranked after evaluating the evidence within each domain. For comparison to this
ranking process, a large-scale candidate gene hypothesis was also tested by including genes with Gene Ontology terms related to neurodevelopment. Subsequently, genotypes of 3725 SNPs in 167 genes from a custom Illumina iSelect array were used to evaluate the top ranked vs. hypothesis selected genes. Seventy-three genes were both highly ranked and involved in neurodevelopment (category 1) while 42 and 52 genes were exclusive to neurodevelopment (category 2) or highly ranked (category 3), respectively. The most significant associations were observed in genes PRKG1, PRKCE, and CNTN4 but no individual SNPs were significant after correction for multiple testing. Comparison of the approaches showed an excess of significant tests using the hypothesis-driven neurodevelopment category. Random selection of similar sized genes from two independent genome-wide association studies (GWAS) of schizophrenia showed the excess was unlikely by chance. In a further meta-analysis of three GWAS datasets, four candidate SNPs reached nominal significance. Although gene ranking using integrated sources of prior information did not enrich for significant results in the current experiment, gene selection using an a priori hypothesis (neurodevelopment) was superior to random selection. As such, further development of gene ranking strategies using more carefully selected sources of information is warranted.