Large Scale Application of Neural Network Based Semantic Role Labeling for Automated Relation Extraction from Biomedical Texts

To reduce the increasing amount of time spent on literature search in the life sciences, several methods for automated knowledge extraction have been developed. Co-occurrence based approaches can deal with large text corpora like MEDLINE in an acceptable time but are not able to extract any specific type of semantic relation. Semantic relation extraction methods based on syntax trees, on the other hand, are computationally expensive and the interpretation of the generated trees is difficult. Several natural language processing (NLP) approaches for the biomedical domain exist focusing specifically on the detection of a limited set of relation types. For systems biology, generic approaches for the detection of a multitude of relation types which in addition are able to process large text corpora are needed but the number of systems meeting both requirements is very limited. We introduce the use of SENNA (“Semantic Extraction using a Neural Network Architecture”), a fast and accurate neural network based Semantic Role Labeling (SRL) program, for the large scale extraction of semantic relations from the biomedical literature. A comparison of processing times of SENNA and other SRL systems or syntactical parsers used in the biomedical domain revealed that SENNA is the fastest Proposition Bank (PropBank) conforming SRL program currently available. 89 million biomedical sentences were tagged with SENNA on a 100 node cluster within three days. The accuracy of the presented relation extraction approach was evaluated on two test sets of annotated sentences resulting in precision/recall values of 0.71/0.43. We show that the accuracy as well as processing speed of the proposed semantic relation extraction approach is sufficient for its large scale application on biomedical text. The proposed approach is highly generalizable regarding the supported relation types and appears to be especially suited for general-purpose, broad-scale text mining systems. The presented approach bridges the gap between fast, cooccurrence-based approaches lacking semantic relations and highly specialized and computationally demanding NLP approaches.     

Large Neural Network Simulations on Multiple Hardware Platforms

To efficiently simulate very large networks of interconnected neurons, particular consideration has to be given to the computer architecture being used. This article presents techniques for implementing simulators for large neural networks on a number of different computer architectures. The neuronal simulation task and the computer architectures of interest are first characterized, and the potential bottlenecks are highlighted. Then we describe the experience gained from adapting an existing simulator, SWIM, to two very different architectures–vector computers and multiprocessor workstations. This work lead to the implementation of a new simulation library, SPLIT, designed to allow efficient simulation of large networks on several architectures. Different computer architectures put different demands on the organization of both data structures and computations. Strict separation of such architecture considerations from the neuronal models and other simulation aspects makes it possible to construct both portable and extendible code.     

Large Scale Molecular Dynamics on Parallel Computers using the Link-cell Algorithm

Parallel computers offer a more cost-effective route to high performance computing than traditional single processor machines. Software for such machines is still in its infancy and they are often much more difficult to program than sequential machines. In addition many of the algorithms which are successful with sequential and vector processors are no longer appropriate. Both the force calculation and integration steps of molecular dynamics are parallel in nature and for that reason we have developed a parallel algorithm based on the link cell technique. This method is particularly efficient when the range of intermolecular potential is much smaller than the dimensions of the simulation box. The details of the algorithm are presented for systems of atoms in two and three dimensions using a number of decompositions into sub-units. The algorithm has been tested on an Intel iPSC/2 and a Cray X-MP/416 and the results are presented for simulations of up to 2 · 10⁶ atoms.     

Modelling Population Dynamics in Realistic Landscapes with Linear Elements: A Mechanistic-Statistical Reaction-Diffusion Approach

We propose and develop a general approach based on reaction-diffusion equations for modelling a species dynamics in a realistic two-dimensional (2D) landscape crossed by linear one-dimensional (1D) corridors, such as roads, hedgerows or rivers. Our approach is based on a hybrid “2D/1D model”, i.e, a system of 2D and 1D reaction-diffusion equations with homogeneous coefficients, in which each equation describes the population dynamics in a given 2D or 1D element of the landscape. Using the example of the range expansion of the tiger mosquito Aedes albopictus in France and its main highways as 1D corridors, we show that the model can be fitted to realistic observation data. We develop a mechanistic-statistical approach, based on the coupling between a model of population dynamics and a probabilistic model of the observation process. This allows us to bridge the gap between the data (3 levels of infestation, at the scale of a French department) and the output of the model (population densities at each point of the landscape), and to estimate the model parameter values using a maximum-likelihood approach. Using classical model comparison criteria, we obtain a better fit and a better predictive power with the 2D/1D model than with a standard homogeneous reaction-diffusion model. This shows the potential importance of taking into account the effect of the corridors (highways in the present case) on species dynamics. With regard to the particular case of A. albopictus, the conclusion that highways played an important role in species range expansion in mainland France is consistent with recent findings from the literature.     

Large Impact of Eurasian Lynx Predation on Roe Deer Population Dynamics

The effects of predation on ungulate populations depend on several factors. One of the most important factors is the proportion of predation that is additive or compensatory respectively to other mortality in the prey, i.e., the relative effect of top-down and bottom-up processes. We estimated Eurasian lynx (Lynx lynx) kill rate on roe deer (Capreolus capreolus) using radio-collared lynx. Kill rate was strongly affected by lynx social status. For males it was 4.85 ± 1.30 S.E. roe deer per 30 days, for females with kittens 6.23 ± 0.83 S.E. and for solitary females 2.71 ± 0.47 S.E. We found very weak support for effects of prey density (both for Type I (linear) and Type II (non-linear) functional responses) and of season (winter, summer) on lynx kill rate. Additionally, we analysed the growth rate in a roe deer population from 1985 to 2005 in an area, which lynx naturally re-colonized in 1996. The annual roe deer growth rate was lower after lynx re-colonized the study area, but it was also negatively influenced by roe deer density. Before lynx colonized the area roe deer growth rate was λ = 1.079 (± 0.061 S.E.), while after lynx re-colonization it was λ = 0.94 (± 0.051 S.E.). Thus, the growth rate in the roe deer population decreased by Δλ = 0.14 (± 0.080 S.E.) after lynx re-colonized the study area, which corresponded to the estimated lynx predation rate on roe deer (0.11 ± 0.042 S.E.), suggesting that lynx predation was mainly additive to other mortality in roe deer. To conclude, this study suggests that lynx predation together with density dependent factors both influence the roe deer population dynamics. Thus, both top-down and bottom-up processes operated at the same time in this predator-prey system.     

Nonlinear Dynamics Analysis of a Self-Organizing Recurrent Neural Network: Chaos Waning

Self-organization is thought to play an important role in structuring nervous systems. It frequently arises as a consequence of plasticity mechanisms in neural networks: connectivity determines network dynamics which in turn feed back on network structure through various forms of plasticity. Recently, self-organizing recurrent neural network models (SORNs) have been shown to learn non-trivial structure in their inputs and to reproduce the experimentally observed statistics and fluctuations of synaptic connection strengths in cortex and hippocampus. However, the dynamics in these networks and how they change with network evolution are still poorly understood. Here we investigate the degree of chaos in SORNs by studying how the networks'' self-organization changes their response to small perturbations. We study the effect of perturbations to the excitatory-to-excitatory weight matrix on connection strengths and on unit activities. We find that the network dynamics, characterized by an estimate of the maximum Lyapunov exponent, becomes less chaotic during its self-organization, developing into a regime where only few perturbations become amplified. We also find that due to the mixing of discrete and (quasi-)continuous variables in SORNs, small perturbations to the synaptic weights may become amplified only after a substantial delay, a phenomenon we propose to call deferred chaos.     

A Spike-Timing Pattern Based Neural Network Model for the Study of Memory Dynamics

It is well accepted that the brain''s computation relies on spatiotemporal activity of neural networks. In particular, there is growing evidence of the importance of continuously and precisely timed spiking activity. Therefore, it is important to characterize memory states in terms of spike-timing patterns that give both reliable memory of firing activities and precise memory of firing timings. The relationship between memory states and spike-timing patterns has been studied empirically with large-scale recording of neuron population in recent years. Here, by using a recurrent neural network model with dynamics at two time scales, we construct a dynamical memory network model which embeds both fast neural and synaptic variation and slow learning dynamics. A state vector is proposed to describe memory states in terms of spike-timing patterns of neural population, and a distance measure of state vector is defined to study several important phenomena of memory dynamics: partial memory recall, learning efficiency, learning with correlated stimuli. We show that the distance measure can capture the timing difference of memory states. In addition, we examine the influence of network topology on learning ability, and show that local connections can increase the network''s ability to embed more memory states. Together theses results suggest that the proposed system based on spike-timing patterns gives a productive model for the study of detailed learning and memory dynamics.     

A Large Scale Huntingtin Protein Interaction Network Implicates Rho GTPase Signaling Pathways in Huntington Disease

Huntington disease (HD) is an inherited neurodegenerative disease caused by a CAG expansion in the HTT gene. Using yeast two-hybrid methods, we identified a large set of proteins that interact with huntingtin (HTT)-interacting proteins. This network, composed of HTT-interacting proteins (HIPs) and proteins interacting with these primary nodes, contains 3235 interactions among 2141 highly interconnected proteins. Analysis of functional annotations of these proteins indicates that primary and secondary HIPs are enriched in pathways implicated in HD, including mammalian target of rapamycin, Rho GTPase signaling, and oxidative stress response. To validate roles for HIPs in mutant HTT toxicity, we show that the Rho GTPase signaling components, BAIAP2, EZR, PIK3R1, PAK2, and RAC1, are modifiers of mutant HTT toxicity. We also demonstrate that Htt co-localizes with BAIAP2 in filopodia and that mutant HTT interferes with filopodial dynamics. These data indicate that HTT is involved directly in membrane dynamics, cell attachment, and motility. Furthermore, they implicate dysregulation in these pathways as pathological mechanisms in HD.     

Iterative Elastic 3D-to-2D Alignment Method Using Normal Modes for Studying Structural Dynamics of Large Macromolecular Complexes

1. Download : Download high-res image (504KB)
2. Download : Download full-size image

     

Population Dynamics Modelling in an Hierarchical Arborescent River Network: An Attempt with Salmo trutta

The balance between births and deaths in an age-structured population is strongly influenced by the spatial distribution of sub-populations. Our aim was to describe the demographic process of a fish population in an hierarchical dendritic river network, by taking into account the possible movements of individuals. We tried also to quantify the effect of river network changes (damming or channelling) on the global fish population dynamics. The Salmo trutta life pattern was taken as an example for.We proposed a model which includes the demographic and the migration processes, considering migration fast compared to demography. The population was divided into three age-classes and subdivided into fifteen spatial patches, thus having 45 state variables. Both processes were described by means of constant transfer coefficients, so we were dealing with a linear system of difference equations. The discrete case of the variable aggregation method allowed the study of the system through the dominant elements of a much simpler linear system with only three global variables: the total number of individuals in each age-class.From biological hypothesis on demographic and migratory parameters, we showed that the global population dynamics of fishes is well characterized in the reference river network, and that dams could have stronger effects on the global dynamics than channelling.     

Explaining Spatial Heterogeneity in Population Dynamics and Genetics from Spatial Variation in Resources for a Large Herbivore

Fine-scale spatial variation in genetic relatedness and inbreeding occur across continuous distributions of several populations of vertebrates; however, the basis of observed variation is often left untested. Here we test the hypothesis that prior observations of spatial patterns in genetics for an island population of feral horses (Sable Island, Canada) were the result of spatial variation in population dynamics, itself based in spatial heterogeneity in underlying habitat quality. In order to assess how genetic and population structuring related to habitat, we used hierarchical cluster analysis of water sources and an indicator analysis of the availability of important forage species to identify a longitudinal gradient in habitat quality along the length of Sable Island. We quantify a west-east gradient in access to fresh water and availability of two important food species to horses: sandwort, Honckenya peploides, and beach pea, Lathyrus japonicas. Accordingly, the population clusters into three groups that occupy different island segments (west, central, and east) that vary markedly in their local dynamics. Density, body condition, and survival and reproduction of adult females were highest in the west, followed by central and east areas. These results mirror a previous analysis of genetics, which showed that inbreeding levels are highest in the west (with outbreeding in the east), and that there are significant differences in fixation indices among groups of horses along the length of Sable Island. Our results suggest that inbreeding depression is not an important limiting factor to the horse population. We conclude that where habitat gradients exist, we can anticipate fine-scale heterogeneity in population dynamics and hence genetics.     

The Association between the Levels of Serum Ferritin and Sex Hormones in a Large Scale of Chinese Male Population

Background

The ferritin is an important participant of iron-storage but its regulation and related factors were not well defined. The present objective was to explore the potential association between serum ferritin levels and sex hormones.

Methods

1999 Chinese men in the Fangchenggang Area Male Health and Examination Survey (FAMHES) were recruited in this cross-sectional study. Levels of serum ferritin, total testosterone (free testosterone was calculated from the total one), estradiol and sex hormone-binding protein were detected in venous blood samples. The effects of age, BMI, smoking as well as alcohol consumption were analyzed on ferritin levels, respectively, and then the Pearson’s correlation analysis was used to evaluate the association between ferritin levels and sex hormones adjusting for the above factors.

Results

The age, BMI and alcohol consumption significantly affected serum ferritin levels, but there was no significant difference between smokers and nonsmokers. Ferritin levels were significantly and negatively associated with total testosterone (R = −0.205, P< 0.001), sex hormone-binding protein (R = −0.161, P<0.001) and free testosterone (R = −0.097, P<0.001). After age and alcohol consumption were adjusted, the above associations were still significant (R = −0.200, −0.181 and −0.083, respectively, all P<0.001). However, there was only borderline negative association between ferritin levels and estradiol (adjusted R = −0.039, P = 0.083).

Conclusion

The large scale of epidemic results showed the significantly negative associations between serum ferritin levels and sex hormones, which may provide more clues to explore the potential regulation and biological mechanism of ferritin.     

Structural Validity of the Tonic Immobility Scale in a Population Exposed to Trauma: Evidence from Two Large Brazilian Samples

Background

Tonic Immobility is a temporary state of motor inhibition in situations involving extreme fear. The first scale developed for its assessment was the 10-item Tonic Immobility Scale (TIS). However, there are still few studies on its structural (dimensional) validity. The objective of this study was to reassess the factor structure of the TIS applied to representative samples exposed to general trauma of two Brazilian mega-cities.

Methods

The sample comprised 3,223 participants reporting at least one traumatic experience. In São Paulo (n = 2,148), a Confirmatory Factor Analysis (CFA) first tested the originally proposed two-dimensional structure. This was followed by sequential Exploratory Structural Equation Models to identify the best fitting model, and subsequently tested in Rio de Janeiro (n = 1,075) via CFA. Alternative reduced versions were further explored using the aggregate sample. Model-based Item Response Theory (IRT) location parameters were also investigated.

Results

An absence of factor-based convergent and discriminant validity rejected the original proposition. However, the one-dimensional structure still held several residual correlations. Further exploration indicated the sustainability of reduced versions with seven (alternative A) and six (alternative B) items. Both presented excellent fit and no relevant residual item correlation. According to the IRT location parameters, items in alternative B covered a wider range of the latent trait. The Loevinger''s H scalability coefficients underscored this pattern.

Conclusions

The original model did not hold. A one-factor solution was the most tenable in both large samples, but with significant item residual correlations, indicating that content redundancies persisted. Further reduced and simplified versions of the TIS proved promising. Although studies are yet to be carried out in other settings, it is the authors'' impression that the restricted versions of the TIS are already apt for use in epidemiologic studies since the pros tend to outweigh the cons (as outlined in the Discussion section).     

The Genealogical Population Dynamics of HIV-1 in a Large Transmission Chain: Bridging within and among Host Evolutionary Rates

Transmission lies at the interface of human immunodeficiency virus type 1 (HIV-1) evolution within and among hosts and separates distinct selective pressures that impose differences in both the mode of diversification and the tempo of evolution. In the absence of comprehensive direct comparative analyses of the evolutionary processes at different biological scales, our understanding of how fast within-host HIV-1 evolutionary rates translate to lower rates at the between host level remains incomplete. Here, we address this by analyzing pol and env data from a large HIV-1 subtype C transmission chain for which both the timing and the direction is known for most transmission events. To this purpose, we develop a new transmission model in a Bayesian genealogical inference framework and demonstrate how to constrain the viral evolutionary history to be compatible with the transmission history while simultaneously inferring the within-host evolutionary and population dynamics. We show that accommodating a transmission bottleneck affords the best fit our data, but the sparse within-host HIV-1 sampling prevents accurate quantification of the concomitant loss in genetic diversity. We draw inference under the transmission model to estimate HIV-1 evolutionary rates among epidemiologically-related patients and demonstrate that they lie in between fast intra-host rates and lower rates among epidemiologically unrelated individuals infected with HIV subtype C. Using a new molecular clock approach, we quantify and find support for a lower evolutionary rate along branches that accommodate a transmission event or branches that represent the entire backbone of transmitted lineages in our transmission history. Finally, we recover the rate differences at the different biological scales for both synonymous and non-synonymous substitution rates, which is only compatible with the ‘store and retrieve’ hypothesis positing that viruses stored early in latently infected cells preferentially transmit or establish new infections upon reactivation.