PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Graph Neural Networks”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original 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
PubMed2025

IGCN: integrative graph convolution networks for patient level insights and biomarker discovery in multi-omics integration.

MOTIVATION: Developing computational tools for integrative analysis across multiple types of omics data has been of immense importance in cancer molecular biology and precision medicine research. While recent advancements have yielded integrative prediction solutions for multi-omics data, these methods lack a comprehensive and cohesive understanding of the rationale behind their specific predictions. To shed light on personalized medicine and unravel previously unknown characteristics within integrative analysis of multi-omics data, we introduce a novel integrative neural network approach for cancer molecular subtype and biomedical classification applications, named Integrative Graph Convolutional Networks (IGCN). RESULTS: To demonstrate the superiority of IGCN, we compare its performance with other state-of-the-art approaches across different cancer subtype and biomedical classification tasks. Our experimental results show that our proposed model outperforms the state-of-the-art and baseline methods. IGCN identifies which types of omics data receive more emphasis for each patient when predicting a specific class. Additionally, IGCN has the capability to pinpoint significant biomarkers from a range of omics data types. AVAILABILITY AND IMPLEMENTATION: The source code is available at https://github.com/bozdaglab/IGCN.

Humans↗
PubMedDate not supplied

General melting point prediction based on a diverse compound data set and artificial neural networks.

We report the development of a robust and general model for the prediction of melting points. It is based on a diverse data set of 4173 compounds and employs a large number of 2D and 3D descriptors to capture molecular physicochemical and other graph-based properties. Dimensionality reduction is performed by principal component analysis, while a fully connected feed-forward back-propagation artificial neural network is employed for model generation. The melting point is a fundamental physicochemical property of a molecule that is controlled by both single-molecule properties and intermolecular interactions due to packing in the solid state. Thus, it is difficult to predict, and previously only melting point models for clearly defined and smaller compound sets have been developed. Here we derive the first general model that covers a comparatively large and relevant part of organic chemical space. The final model is based on 2D descriptors, which are found to contain more relevant information than the 3D descriptors calculated. Internal random validation of the model achieves a correlation coefficient of R(2) = 0.661 with an average absolute error of 37.6 degrees C. The model is internally consistent with a correlation coefficient of the test set of Q(2) = 0.658 (average absolute error 38.2 degrees C) and a correlation coefficient of the internal validation set of Q(2) = 0.645 (average absolute error 39.8 degrees C). Additional validation was performed on an external drug data set consisting of 277 compounds. On this external data set a correlation coefficient of Q(2) = 0.662 (average absolute error 32.6 degrees C) was achieved, showing ability of the model to generalize. Compared to an earlier model for the prediction of melting points of druglike compounds our model exhibits slightly improved performance, despite the much larger chemical space covered. The remaining model error is due to molecular properties that are not captured using single-molecule based descriptors, namely both inter- and intramolecular interactions and crystal packing, for which examples of and reasons for outliers are given.

Journal Article↗
PubMed1979

Structure and dynamics of neural network oscillators.

Techniques are given to represent oscillating neural networks by asynchronous logical switching networks, and to analyze the oscillating networks using a directed graph called a state transition diagram. Consideration is restricted to network oscillators containing no rhythm determining pacemaker neurons, and no neurons exhibiting self-limiting properties such as post-inhibitory rebound or accumulating refractoriness. In the state transition diagrams, stable oscillations are associated with a particular geometric configuration called a cyclic attractor (the heavy cycle in Fig. 2). We show that given the network connectivity it is possible to predict autonomous dynamic behaviour, as well as behaviour following hyperpolarizing or depolarizing inputs to neurons of the network. Conversely, given information about patterns of firing activity during cycles and transients of neural networks, the network connectivity can be predicted. The theoretical techniques can be used to generate a census of network structures capable of generating stable oscillations. Several representative network oscillators are discussed in the context of previous theoretical and experimental studies of the structure of neural network oscillators. Although the number of theoretically possile network oscillators capable of generating sustained oscillations is very large, the techniques which are given should be useful in the design of experiments capable of distinguishing between equally plausible hypotheses.

Action Potentials↗
PubMed2005

A stochastic population approach to the problem of stable recruitment hierarchies in spiking neural networks.

Synchrony-driven recruitment learning addresses the question of how arbitrary concepts, represented by synchronously active ensembles, may be acquired within a randomly connected static graph of neuron-like elements. Recruitment learning in hierarchies is an inherently unstable process. This paper presents conditions on parameters for a feedforward network to ensure stable recruitment hierarchies. The parameter analysis is conducted by using a stochastic population approach to model a spiking neural network. The resulting network converges to activate a desired number of units at each stage of the hierarchy. The original recruitment method is modified first by increasing feedforward connection density for ensuring sufficient activation, then by incorporating temporally distributed feedforward delays for separating inputs temporally, and finally by limiting excess activation via lateral inhibition. The task of activating a desired number of units from a population is performed similarly to a temporal k-winners-take-all network.

Feedback, Physiological↗
PubMedDate not supplied

Using neural network predicted secondary structure information in automatic protein NMR assignment.

In CAPRI, an automated NMR assignment software package that was developed in our laboratory, both chemical shift values and coupling topologies of spin patterns are used in a procedure for amino acids recognition. By using a knowledge base of chemical shift distributions of the 20 amino acid types, fuzzy mathematics, and pattern recognition theory, the spin coupling topological graphs are mapped onto specific amino acid residues. In this work, we investigated the feasibility of using secondary structure information of proteins as predicted by neural networks in the automated NMR assignment. As the 1H and 13C chemical shifts of proteins are known to correlate to their secondary structures, secondary structure information is useful in improving the amino acid recognition. In this study, the secondary structures of proteins predicted by the PHD protein server and our own trained neural networks are used in the amino acid type recognition. The results show that the predicted secondary structure information can help to improve the accuracy of the amino acid recognition.

Algorithms↗
PubMed2003

Duplication models for biological networks.

Are biological networks different from other large complex networks? Both large biological and nonbiological networks exhibit power-law graphs (number of nodes with degree k, N(k) approximately k(-beta)), yet the exponents, beta, fall into different ranges. This may be because duplication of the information in the genome is a dominant evolutionary force in shaping biological networks (like gene regulatory networks and protein-protein interaction networks) and is fundamentally different from the mechanisms thought to dominate the growth of most nonbiological networks (such as the Internet). The preferential choice models used for nonbiological networks like web graphs can only produce power-law graphs with exponents greater than 2. We use combinatorial probabilistic methods to examine the evolution of graphs by node duplication processes and derive exact analytical relationships between the exponent of the power law and the parameters of the model. Both full duplication of nodes (with all their connections) as well as partial duplication (with only some connections) are analyzed. We demonstrate that partial duplication can produce power-law graphs with exponents less than 2, consistent with current data on biological networks. The power-law exponent for large graphs depends only on the growth process, not on the starting graph.

Internet↗
PubMed2005

Causal connectivity of evolved neural networks during behavior.

To show how causal interactions in neural dynamics are modulated by behavior, it is valuable to analyze these interactions without perturbing or lesioning the neural mechanism. This paper proposes a method, based on a graph-theoretic extension of vector autoregressive modeling and 'Granger causality,' for characterizing causal interactions generated within intact neural mechanisms. This method, called 'causal connectivity analysis' is illustrated via model neural networks optimized for controlling target fixation in a simulated head-eye system, in which the structure of the environment can be experimentally varied. Causal connectivity analysis of this model yields novel insights into neural mechanisms underlying sensorimotor coordination. In contrast to networks supporting comparatively simple behavior, networks supporting rich adaptive behavior show a higher density of causal interactions, as well as a stronger causal flow from sensory inputs to motor outputs. They also show different arrangements of 'causal sources' and 'causal sinks': nodes that differentially affect, or are affected by, the remainder of the network. Finally, analysis of causal connectivity can predict the functional consequences of network lesions. These results suggest that causal connectivity analysis may have useful applications in the analysis of neural dynamics.

Aging↗
PubMed2025

Exploring the transmission of cognitive task information through optimal brain pathways.

Understanding the large-scale information processing that underlies complex human cognition is the central goal of cognitive neuroscience. While emerging activity flow models demonstrate that cognitive task information is transferred by interregional functional or structural connectivity, graph-theory-based models typically assume that neural communication occurs via the shortest path of brain networks. However, whether the shortest path is the optimal route for empirical cognitive information transmission remains unclear. Based on a large-scale activity flow mapping framework, we found that the performance of activity flow prediction with the shortest path was significantly lower than that with the direct path. The shortest path routing was superior to other network communication strategies, including search information, path ensembles, and navigation. Intriguingly, the shortest path outperformed the direct path in activity flow prediction when the physical distance constraint and asymmetric routing contribution were simultaneously considered. This study not only challenges the shortest path assumption through empirical network models but also suggests that cognitive task information routing is constrained by the spatial and functional embedding of the brain network.

Humans↗
PubMed2005

Evolution and similarity evaluation of protein structures in contact map space.

Prediction of fold from amino acid sequence of a protein has been an active area of research in the past few years, but the limited accuracy of existing techniques emphasizes the need to develop newer approaches to tackle this task. In this study, we use contact map prediction as an intermediate step in fold prediction from sequence. Contact map is a reduced graph-theoretic representation of proteins that models the local and global inter-residue contacts in the structure. We start with a population of random contact maps for the protein sequence and "evolve" the population to a "high-feasibility" configuration using a genetic algorithm. A neural network is employed to assess the feasibility of contact maps based on their 4 physically relevant properties. We also introduce 5 parameters, based on algebraic graph theory and physical considerations, that can be used to judge the structural similarity between proteins through contact maps. To predict the fold of a given amino acid sequence, we predict a contact map that will sufficiently approximate the structure of the corresponding protein. Then we assess the similarity of this contact map with the representative contact map of each fold; the fold that corresponds to the closest match is our predicted fold for the input sequence. We have found that our feasibility measure is able to differentiate between feasible and infeasible contact maps. Further, this novel approach is able to predict the folds from sequences significantly better than a random predictor.

Amino Acid Sequence↗
PubMed2004

An annealed chaotic maximum neural network for bipartite subgraph problem.

In this paper, based on maximum neural network, we propose a new parallel algorithm that can help the maximum neural network escape from local minima by including a transient chaotic neurodynamics for bipartite subgraph problem. The goal of the bipartite subgraph problem, which is an NP- complete problem, is to remove the minimum number of edges in a given graph such that the remaining graph is a bipartite graph. Lee et al. presented a parallel algorithm using the maximum neural model (winner-take-all neuron model) for this NP- complete problem. The maximum neural model always guarantees a valid solution and greatly reduces the search space without a burden on the parameter-tuning. However, the model has a tendency to converge to a local minimum easily because it is based on the steepest descent method. By adding a negative self-feedback to the maximum neural network, we proposed a new parallel algorithm that introduces richer and more flexible chaotic dynamics and can prevent the network from getting stuck at local minima. After the chaotic dynamics vanishes, the proposed algorithm is then fundamentally reined by the gradient descent dynamics and usually converges to a stable equilibrium point. The proposed algorithm has the advantages of both the maximum neural network and the chaotic neurodynamics. A large number of instances have been simulated to verify the proposed algorithm. The simulation results show that our algorithm finds the optimum or near-optimum solution for the bipartite subgraph problem superior to that of the best existing parallel algorithms.

Algorithms↗
PubMedDate not supplied

eLoom and Flatland: specification, simulation and visualization engines for the study of arbitrary hierarchical neural architectures.

eLoom is an open source graph simulation software tool, developed at the University of New Mexico (UNM), that enables users to specify and simulate neural network models. Its specification language and libraries enables users to construct and simulate arbitrary, potentially hierarchical network structures on serial and parallel processing systems. In addition, eLoom is integrated with UNM's Flatland, an open source virtual environments development tool to provide real-time visualizations of the network structure and activity. Visualization is a useful method for understanding both learning and computation in artificial neural networks. Through 3D animated pictorially representations of the state and flow of information in the network, a better understanding of network functionality is achieved. ART-1, LAPART-II, MLP, and SOM neural networks are presented to illustrate eLoom and Flatland's capabilities.

Computer Simulation↗
PubMed2002

A bi-recursive neural network architecture for the prediction of protein coarse contact maps.

Prediction of contact maps may be seen as a strategic step towards the solution of fundamental open problems in structural genomics. In this paper we focus on coarse grained maps that describe the spatial neighborhood relation between secondary structure elements (helices, strands, and coils) of a protein. We introduce a new machine learning approach for scoring candidate contact maps. The method combines a specialized noncausal recursive connectionist architecture and a heuristic graph search algorithm. The network is trained using candidate graphs generated during search. We show how the process of selecting and generating training examples is important for tuning the precision of the predictor.

Algorithms↗
PubMedDate not supplied

Analysis of the internal representations developed by neural networks for structures applied to quantitative structure--activity relationship studies of benzodiazepines.

An application of recursive cascade correlation (CC) neural networks to quantitative structure-activity relationship (QSAR) studies is presented, with emphasis on the study of the internal representations developed by the neural networks. Recursive CC is a neural network model recently proposed for the processing of structured data. It allows the direct handling of chemical compounds as labeled ordered directed graphs, and constitutes a novel approach to QSAR. The adopted representation of molecular structure captures, in a quite general and flexible way, significant topological aspects and chemical functionalities for each specific class of molecules showing a particular chemical reactivity or biological activity. A class of 1,4-benzodiazepin-2-ones is analyzed by the proposed approach. It compares favorably versus the traditional QSAR treatment based on equations. To show the ability of the model in capturing most of the structural features that account for the biological activity, the internal representations developed by the networks are analyzed by principal component analysis. This analysis shows that the networks are able to discover relevant structural features just on the basis of the association between the molecular morphology and the target property (affinity).

Benzodiazepines↗
PubMed2002

Automatic recognition of cortical sulci of the human brain using a congregation of neural networks.

This paper describes a complete system allowing automatic recognition of the main sulci of the human cortex. This system relies on a preprocessing of magnetic resonance images leading to abstract structural representations of the cortical folding patterns. The representation nodes are cortical folds, which are given a sulcus name by a contextual pattern recognition method. This method can be interpreted as a graph matching approach, which is driven by the minimization of a global function made up of local potentials. Each potential is a measure of the likelihood of the labelling of a restricted area. This potential is given by a multi-layer perceptron trained on a learning database. A base of 26 brains manually labelled by a neuroanatomist is used to validate our approach. The whole system developed for the right hemisphere is made up of 265 neural networks. The mean recognition rate is 86% for the learning base and 76% for a generalization base, which is very satisfying considering the current weak understanding of the variability of the cortical folding patterns.

Artificial Intelligence↗
PubMed2003

An efficient approximation algorithm for finding a maximum clique using Hopfield network learning.

In this article, we present a solution to the maximum clique problem using a gradient-ascent learning algorithm of the Hopfield neural network. This method provides a near-optimum parallel algorithm for finding a maximum clique. To do this, we use the Hopfield neural network to generate a near-maximum clique and then modify weights in a gradient-ascent direction to allow the network to escape from the state of near-maximum clique to maximum clique or better. The proposed parallel algorithm is tested on two types of random graphs and some benchmark graphs from the Center for Discrete Mathematics and Theoretical Computer Science (DIMACS). The simulation results show that the proposed learning algorithm can find good solutions in reasonable computation time.

Algorithms↗
PubMed2005

Geometric diffusions for the analysis of data from sensor networks.

Harmonic analysis on manifolds and graphs has recently led to mathematical developments in the field of data analysis. The resulting new tools can be used to compress and analyze large and complex data sets, such as those derived from sensor networks or neuronal activity datasets, obtained in the laboratory or through computer modeling. The nature of the algorithms (based on diffusion maps and connectivity strengths on graphs) possesses a certain analogy with neural information processing, and has the potential to provide inspiration for modeling and understanding biological organization in perception and memory formation.

Algorithms↗
PubMed2004

A hopfield network learning method for bipartite subgraph problem.

In this paper, we present a gradient ascent learning method of the Hopfield neural network for bipartite subgraph problem. The method is intended to provide a near-optimum parallel algorithm for solving the bipartite subgraph problem. To do this we use the Hopfield neural network to get a near-maximum bipartite subgraph, and increase the energy by modifying weights in a gradient ascent direction of the energy to help the network escape from the state of the near-maximum bipartite subgraph to the state of the maximum bipartite subgraph or better one. A large number of instances are simulated to verify the proposed method with the simulation results showing that the solution quality is superior to that of best existing parallel algorithm. We also test the learning method on total coloring problem. The simulation results show that our method finds optimal solution in every test graph.

Algorithms↗