Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Ensemble methods”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4Linked to original sources
PubMed2000

A single-molecule study of RNA catalysis and folding.

Using fluorescence microscopy, we studied the catalysis by and folding of individual Tetrahymena thermophila ribozyme molecules. The dye-labeled and surface-immobilized ribozymes used were shown to be functionally indistinguishable from the unmodified free ribozyme in solution. A reversible local folding step in which a duplex docks and undocks from the ribozyme core was observed directly in single-molecule time trajectories, allowing the determination of the rate constants and characterization of the transition state. A rarely populated docked state, not measurable by ensemble methods, was observed. In the overall folding process, intermediate folding states and multiple folding pathways were observed. In addition to observing previously established folding pathways, a pathway with an observed folding rate constant of 1 per second was discovered. These results establish single-molecule fluorescence as a powerful tool for examining RNA folding.

Animals↗
PubMed2005

Propensity scores: an introduction and experimental test.

Propensity score analysis is a relatively recent statistical innovation that is useful in the analysis of data from quasi-experiments. The goal of propensity score analysis is to balance two non-equivalent groups on observed covariates to get more accurate estimates of the effects of a treatment on which the two groups differ. This article presents a general introduction to propensity score analysis, provides an example using data from a quasi-experiment compared to a benchmark randomized experiment, offers practical advice about how to do such analyses, and discusses some limitations of the approach. It also presents the first detailed instructions to appear in the literature on how to use classification tree analysis and bagging for classification trees in the construction of propensity scores. The latter two examples serve as an introduction for researchers interested in computing propensity scores using more complex classification algorithms known as ensemble methods.

Algorithms↗
PubMed2026

CLASPP: A unified model for predicting post-translational modifications.

Post-Translational Modifications (PTMs) are a fundamental mechanism for regulating cellular pathways and increasing the functional diversity of the proteome. Accurately predicting the PTM types that are likely to occur at a given site in the primary sequence is a key challenge in functional proteomics. Existing PTM prediction models predominantly focus on either single PTM types or employ ensemble methods that combine multiple models to predict different PTM types. This fragmentation is largely driven by the vast imbalance in data availability across PTM types, making it difficult to predict multiple PTM types with a single model. To address this limitation, we present the Contrastively Learned Attention-based Stratified PTM Predictor (CLASPP), a unified PTM prediction model. CLASPP addresses imbalance challenges by leveraging unsupervised clustering-based undersampling and a novel contrastive learning framework tailored to PTM data. Additionally, our hierarchical data organization and curation are shown to improve CLASPP's performance by balancing the representation of individual PTM types and provides a standardized dataset to train and validate future model designs. Drawing inspiration from advancements in image and natural language processing, the CLASPP model employs a multi-stage training strategy and a high-quality, curated training dataset to improve PTM prediction performance. To uncover what is learned during the contrastive learning stage, the CLASPP model is shown to distinguish known protein kinase substrate specificity profiles as a form of explainability. Finally, we evaluate the application of CLASPP in predicting PTMs in different model organisms and experimentally validated ubiquitination sites in the understudied DCLK3 kinase. Overall, CLASPP represents a unified model for PTM prediction that addresses key bottlenecks in data imbalance and offers new strategies for biological data curation, thereby improving PTM-type prediction performance across diverse organisms.

Protein Processing, Post-Translational↗
PubMed2004

Abductive network committees for improved classification of medical data.

OBJECTIVES: To introduce abductive network classifier committees as an ensemble method for improving classification accuracy in medical diagnosis. While neural networks allow many ways to introduce enough diversity among member models to improve performance when forming a committee, the self-organizing, automatic-stopping nature, and learning approach used by abductive networks are not very conducive for this purpose. We explore ways of over-coming this limitation and demonstrate improved classification on three standard medical datasets. METHODS: Two standard 2-class medical datasets (Pima Indians Diabetes and Heart Disease) and a 6-class dataset (Dermatology) were used to investigate ways of training abductive networks with adequate independence, as well as methods of combining their outputs to form a network that improves performance beyond that of single models. RESULTS: Two- or three-member committees of models trained on completely or partially different subsets of training data and using simple output combination methods achieve improvements between 2 and 5 percentage points in the classification accuracy over the best single model developed using the full training set. CONCLUSIONS: Varying model complexity alone gives abductive network models that are too correlated to ensure enough diversity for forming a useful committee. Diversity achieved through training member networks on independent subsets of the training data outweighs limitations of the smaller training set for each, resulting in net gain in committee performance. As such models train faster and can be trained in parallel, this can also speed up classifier development.

Data Collection↗
PubMed2004

Determination of fluid--solid transitions in model protein solutions using the histogram reweighting method and expanded ensemble simulations.

Protein crystallization conditions are usually identified by empirical screening methods because of the complexity of the process, such as the existence of nonequilibrium phases and the different crystal forms that may result from changes in solution conditions. Here the crystallization of a model protein is studied using computer simulation. The model consists of spheres that have both an isotropic interaction of short range and anisotropic interactions between patch-antipatch pairs. The free energy of a protein crystal is calculated using expanded ensemble simulations of the Einstein crystal, and NpT-Monte Carlo simulations with histogram reweighting are used to determine the fluid-solid coexistence. The histogram reweighting method is also used to trace out the complete coexistence curve, including multiple crystal phases, with varying reduced temperature, which corresponds to changing solution conditions. At a patch-antipatch interaction strength five times that of the isotropic interaction, the protein molecules form a stable simple cubic structure near room temperature, whereas an orientationally disordered face-centered-cubic structure is favored at higher temperatures. The anisotropic attractions also lead to a weak first-order transition between orientationally disordered and ordered face-centered-cubic structures at low temperature, although this transition is metastable. A complete phase diagram, including a fluid phase, three solid phases, and two triple points, is found for the six-patch protein model. A 12-patch protein model, consistent with the face-centered-cubic structure, leads to greater thermodynamic stability of the ordered phase. Metastable liquid-liquid phase equilibria for isotropic models with varying attraction tails are also predicted from Gibbs ensemble simulations.

Crystallization↗
PubMed2004

Dual control cell reaction ensemble molecular dynamics: a method for simulations of reactions and adsorption in porous materials.

We present a simulation tool to study fluid mixtures that are simultaneously chemically reacting and adsorbing in a porous material. The method is a combination of the reaction ensemble Monte Carlo method and the dual control volume grand canonical molecular dynamics technique. The method, termed the dual control cell reaction ensemble molecular dynamics method, allows for the calculation of both equilibrium and nonequilibrium transport properties in porous materials such as diffusion coefficients, permeability, and mass flux. Control cells, which are in direct physical contact with the porous solid, are used to maintain the desired reaction and flow conditions for the system. The simulation setup closely mimics an actual experimental system in which the thermodynamic and flow parameters are precisely controlled. We present an application of the method to the dry reforming of methane reaction within a nanoscale reactor model in the presence of a semipermeable membrane that was modeled as a porous material similar to silicalite. We studied the effects of the membrane structure and porosity on the reaction species permeability by considering three different membrane models. We also studied the effects of an imposed pressure gradient across the membrane on the mass flux of the reaction species. Conversion of syngas (H2/CO) increased significantly in all the nanoscale membrane reactor models considered. A brief discussion of further potential applications is also presented.

Journal Article↗
PubMed2004

Augmented moment method for stochastic ensembles with delayed couplings. I. Langevin model.

By employing a semianalytical dynamical mean-field approximation theory previously proposed by the author [H. Hasegawa, Phys. Rev. E 67, 041903 (2003)], we have developed an augmented moment method (AMM) in order to discuss dynamics of an N -unit ensemble described by Langevin equations with delays. In an AMM, original N -dimensional stochastic delay differential equations (SDDEs) are transformed to infinite-dimensional deterministic DEs for means and correlations of local as well as global variables. Infinite-order DEs arising from the non-Markovian property of SDDE, are terminated at the finite level m in the level-m AMM (AMMm), which yields (3+m)-dimensional deterministic DEs. Model calculations have been made for linear and nonlinear Langevin models. The stationary solution of AMM for the linear Langevin model with N=1 is nicely compared to the exact result. In the nonlinear Langevin ensemble, the synchronization is shown to be enhanced near the transition point between the oscillating and nonoscillating states. Results calculated by AMM6 are in good agreement with those obtained by direct simulations.

Action Potentials↗
PubMed2004

Augmented moment method for stochastic ensembles with delayed couplings. II. FitzHugh-Nagumo model.

Dynamics of FitzHugh-Nagumo (FN) neuron ensembles with time-delayed couplings subject to white noises, has been studied by using both direct simulations and a semianalytical augmented moment method (AMM) which has been proposed in a preceding paper [H. Hasegawa, Phys. Rev. E 70, 021911 (2004)]. For N-unit FN neuron ensembles, AMM transforms original 2N-dimensional stochastic delay differential equations (SDDEs) to infinite-dimensional deterministic DEs for means and correlation functions of local and global variables. Infinite-order recursive DEs are terminated at the finite level m in the level-m AMM (AMMm), yielding 8(m+1)-dimensional deterministic DEs. When a single spike is applied, the oscillation may be induced if parameters of coupling strength, delay, noise intensity and/or ensemble size are appropriate. Effects of these parameters on the emergence of the oscillation and on the synchronization in FN neuron ensembles have been studied. The synchronization shows the fluctuation-induced enhancement at the transition between nonoscillating and oscillating states. Results calculated by AMM5 are in fairly good agreement with those obtained by direct simulations.

Action Potentials↗
PubMed2001

Dynamical approach to the microcanonical ensemble.

An analytical method to compute thermodynamic properties of a given Hamiltonian system is proposed. This method combines ideas of both dynamical systems and ensemble approaches to thermodynamics, providing de facto a possible alternative to traditional ensemble methods. Thermodynamic properties are extracted from effective motion equations. These equations are obtained by introducing a general variational principle applied to an action averaged over a statistical ensemble of paths defined on the constant energy surface. The method is applied first to the one-dimensional beta-Fermi-Pasta-Ulam chain and to the two-dimensional lattice straight phi(4) model. In both cases, the method gives a good insight of some of their statistical and dynamical properties.

Journal Article↗
PubMed2006

On the use of transition matrix methods with extended ensembles.

Different extended ensemble schemes for non-Boltzmann sampling (NBS) of a selected reaction coordinate lambda were formulated so that they employ (i) "variable" sampling window schemes (that include the "successive umbrella sampling" method) to comprehensibly explore the lambda domain and (ii) transition matrix methods to iteratively obtain the underlying free-energy eta landscape (or "importance" weights) associated with lambda. The connection between "acceptance ratio" and transition matrix methods was first established to form the basis of the approach for estimating eta(lambda). The validity and performance of the different NBS schemes were then assessed using as lambda coordinate the configurational energy of the Lennard-Jones fluid. For the cases studied, it was found that the convergence rate in the estimation of eta is little affected by the use of data from high-order transitions, while it is noticeably improved by the use of a broader window of sampling in the variable window methods. Finally, it is shown how an "elastic" window of sampling can be used to effectively enact (nonuniform) preferential sampling over the lambda domain, and how to stitch the weights from separate one-dimensional NBS runs to produce a eta surface over a two-dimensional domain.

Journal Article↗
PubMed2005

Molecular simulations of aqueous electrolyte solubility: 1. The expanded-ensemble osmotic molecular dynamics method for the solution phase.

We have developed a molecular-level simulation technique called the expanded-ensemble osmotic molecular dynamics (EEOMD) method, for studying electrolyte solution systems. The EEOMD method performs simulations at a fixed number of solvent molecules, pressure, temperature, and overall electrolyte chemical potential. The method combines elements of constant pressure-constant temperature molecular dynamics and expanded-ensemble grand canonical Monte Carlo. The simulated electrolyte solution systems contain, in addition to solvent molecules, full and fractional ions and undissociated electrolyte molecular units. The fractional particles are coupled to the system via a coupling parameter that varies between 0 (no interaction between the fractional particle and the other particles in the system) and 1 (full interaction between the fractional particle and the other particles in the system). The time evolution of the system is governed by the constant pressure-constant temperature equations of motion and accompanied by random changes in the coupling parameter. The coupling-parameter changes are accepted with a probability derived from the expanded-ensemble osmotic partition function corresponding to the prescribed electrolyte chemical potential. The coupling-parameter changes mimic insertion/deletion of particles as in a crude grand canonical Monte Carlo simulation; if the coupling parameter becomes 0, the fractional particles disappear from the system, and as the coupling parameter reaches unity, the fractional particles become full particles. The method is demonstrated for a model of NaCl in water at ambient conditions. To test our approach, we first determine the chemical potential of NaCl in water by the thermodynamic integration technique and by the expanded-ensemble method. Then, we carry out EEOMD simulations for different specified values of the overall NaCl chemical potential and measure the concentration of ions resulting from the simulations. Both computations give consistent results, validating the EEOMD methodology.

Journal Article↗
PubMed2005

Face membership authentication using SVM classification tree generated by membership-based LLE data partition.

This paper presents a new membership authentication method by face classification using a support vector machine (SVM) classification tree, in which the size of membership group and the members in the membership group can be changed dynamically. Unlike our previous SVM ensemble-based method, which performed only one face classification in the whole feature space, the proposed method employed a divide and conquer strategy that first performs a recursive data partition by membership-based locally linear embedding (LLE) data clustering, then does the SVM classification in each partitioned feature subset. Our experimental results show that the proposed SVM tree not only keeps the good properties that the SVM ensemble method has, such as a good authentication accuracy and the robustness to the change of members, but also has a considerable improvement on the stability under the change of membership group size.

Decision Trees↗
PubMed1987

Determination of cell membrane permeability in concentrated cell ensembles.

The method of volume averaging is used to analyze the process of diffusion in concentrated cell ensembles in which significant resistance to mass transfer is caused by the cellular membrane. A general closure scheme is given that allows for direct theoretical prediction of effective diffusivities for any cellular geometry. Numerical results are presented for the classical parallelepiped arrangement used to model cellular systems, and these results are used in conjunction with experimental studies of concentrated cell ensembles to determine membrane permeabilities for solute diffusion in several cellular systems. Membrane permeabilities are compared with predictions from other models of diffusion in cellular systems.

Animals↗
PubMed1996

Visual evoked potential enhancement by an artificial neural network filter.

The application of an artificial neural network filter (ANNF) to estimate the visual evoked potential (VEP) is presented. VEP is the gross electrical response of the brain to visual stimuli. Due to the low SNR, it is difficult to extract response from individual stimulus trials. The ANNF we used estimates the deterministic component of the signal and removes the noise uncorrelated with the stimulus, even when the noise is colored. The ANNF is trained through back-error propagation with a data set consisting of a training signal and a target signal. The training signal is the raw VEP from a single trial having a SNR of about -5 dB, while the target signal has a higher SNR which is achieved by ensemble averaging 100 stimulus trials. Simulated signals were generated to test the performance of the ANNF. Results show that the ANNF could greatly enhance the SNR of the VEP to single visual stimulus. Thus the total number of ensembles is reduced. In clinical applications, the traditional ensemble averaging method requires a hundred ensembles to determine the VEP. When ANNF is used, about 20 ensembles are sufficient for the same purpose.

Adolescent↗
PubMed1998

The ensemble-averaged impedance cardiogram: an evaluation of scoring methods and interrater reliability.

The interrater reliability and concurrent validity of two methods of scoring the ensemble-averaged impedance cardiogram were evaluated. Impedance cardiographic and electrocardiographic signals were recorded from 40 undergraduate men and women during a baseline rest period and a vocal mental arithmetic task period. Recordings were scored by four rates using a conventional method, involving ensemble averaging after careful editing of beat-to-beat waveforms, and a streamlined method, involving ensemble averaging without beat-to-beat editing. Intraclass correlations for interrater reliability exceeded .92, whereas intraclass correlations for concurrent validity exceeded .97, indicating excellent agreement between rates and scoring methods for all cardiac measures. The streamlined method was significantly faster than the conventional method. The results indicate that variations in beat-to-beat editing do not constitute a serious source of error in the ensemble-averaged impedance cardiogram and support the interrater reliability and concurrent validity of the two scoring methods.

Adult↗
PubMed2004

Granular contact force density of states and entropy in a modified Edwards ensemble.

A method has been found to analyze Edwards' granular contact force probability functional for a special case. As a result, the granular contact force probability density functions are obtained from first principles for this case. The results are in excellent agreement with the experimental and simulation data. The derivation assumes Edwards' flat measure--a density of states (DOS) that is uniform within the metastable regions of phase space. The enabling assumption, supported by physical arguments and empirical evidence, is that correlating information is not significantly recursive through loops in the packing. Maximizing a state-counting entropy results in a transport equation that can be solved numerically. For the present this has been done using the "mean-structure approximation," projecting the DOS across all angular coordinates to more clearly identify its predominant nonuniformities. These features are (1) the grain factor Psi related to grain stability and strong correlation between the contact forces on the same grain and (2) the structure factor Upsilon related to Newton's third law and strong correlation between neighboring grains.

Journal Article↗