蛋白质二级结构预测的人工神经网络方法研究

本文比较了五种神经网络方法预测蛋白质二级结构的准确率,并做出初步评价。五种神经网络分别是:误差反传前向网络(BP),径向基函数网络(RBF),广义回归神经网络(GRNN),串并联叠层网络(CF),Elman网络(ELM)。结果显示:GRNN的预测准确率达85.7%,优于其它网络。本文还讨论了训练集样本数及参数的优化对GRNN预测准确率的影响。     

Theory of relaxation spectra and dielectric relaxation of rigid rodlike particles incorporated in a polymer network

The theory of molecular mobility and relaxation spectra is developed for rodlike particles embedded in a polymer network with allowance for the involvement of the particles in collective network dynamics through topological entanglements with network fragments. A regular cubic coarse-grained network model is used, where the motion of junctions describes the mobility of large fragments (domains) of the initial network with a size equal to the distance between adjacent rodlike particles. The involvement of the rods in collective network dynamics is taken into account by introducing an effective quasi-elastic potential acting between the rods and junctions of the coarse-grained network and preventing long-distance diffusion of the embedded particles. The viscoelastic parameters of the coarse-grained (“renormalized”) network are functions of the viscoelastic characteristics of the initial network. The relaxation time spectra are calculated as well as the frequency dependences of the dielectric loss factor of the embedded particles that possess a permanent dipole moment directed along the major axis of each rod. Depending on the ratio between the viscoelastic characteristics of the rods and the network, the frequency dependence of the dielectric loss factor may have two maxima. The high-frequency maximum corresponds to local orientational movements of particles at fixed junctions of the coarse-grained network, which correspond to the position of the domain centers in the initial network. The low-frequency maximum corresponds to movements of particles involved in large-scale dynamics of network fragments. The dependence of the dielectric loss factor on the ratio between the viscoelastic parameters of the rods and the network is studied.     

级联神经网络用于单扫描示波极谱信号的测定

将改进小波神经网络与BP神经网络相结合,提出一种新的混级联神经网络结构,用于单扫描示波极谱信号的同时测定。通过对网络结构的优化和网络参数的调整,加快了训练速度,提高了预测的准确度。用该法对邻、间硝基氯苯混合样进行了预测,结果满意。对级联神经网络法与单一BP神经网络法的预测结果进行了比较,表明级联神经网络优于单一BP神经网络。     

Theory of Intrachain Relaxation Spectra for Polymer Networks Possessing a Short- or Long-Scale Ordering. Effects of the Nematic Ordering on the Relaxation Spectrum of a Polymer Network with Included Rods

We discuss the relaxation properties of polymer networks possessing either short-scale ordering caused by rigidity of network strands or long-scale liquid crystalline order. The main topics of the paper are the equilibrium and local dynamic properties of a polymer network ordered due to nematic-like interactions of the network segments with included rod-like particles. A simplified three chain network model is used. Lagrange multipliers in the equations of motion of hard rods are replaced by their averaged values. This approximation corresponds to modelling the rod-like particles by elastic Gaussian springs, their mean-square lengths independent of the ordering. Nematic-like interactions between network segments and rods are taken into account in terms of the Maier-Saupe mean-field approximation. Nematic ordering of rods induces ordering of the network segments. Relaxation spectrum of the ordered network splits into two main branches for the parallel and perpendicular components of the chain segments with respect to the director. We calculate the relaxation times of a polymer network as functions of the wave number. The relaxation spectrum of an isotropic network and that of the ordered network with included rods are compared.     

The Interrelation Between the Microphase Structure of Sequential Interpenetrating Polymer Networks and the Crosslinking Density of Penetrating Networks

The microphase structure of single polyurethane (PU) and acrylate networks as well as sequential interpenetrating polymer networks (IPNs), produced by the forming of a PU network in the presence of monomers of a penetrating network, was studied by small- and wide-angle x-ray analysis. It was established that each network component was of a two-phase structure consisting of disordered phase-separated microregions. the higher crosslink density of the acrylate network results in its higher heterogeneity. in IPNs, phase separation of a complex nature is realized: the PU matrix preserves some features of a single network structure, and the second component forms microregions 5-10 nm in size while retaining a certain level of interpenetra-tion of both network components. the microphase structure parameters of such systems are greatly dependent on the crosslink density of the penetrating network. This suggests the influence of a three-dimensional network of chemical bonds on the interdiffusion of branched fragments of the penetrating network and molecular chains of the matrix, one leading to the retardation of phase separation.     

近红外光谱-人工神经网络法用于粉末药品阿斯匹林的非破坏定量分析

研究了应用人工神经网络进行粉末药品的非破坏定量分析。使用阿斯匹林粉末药品的近红外漫反射一阶导数光谱数据建立人工神经网络模型,预测未知样品。讨论了影响网络的各参数,使用了新的网络评价标准－逼近度。     

Classification and identification of mosquito species using artificial neural networks

An artificial neural network method is presented for classification and identification of Anopheles mosquito species based on the internal transcribed spacer2 (ITS2) data of ribosomal DNA string. The method is implemented in two different multi-layered feed-forward neural network model forms, namely, multi-input single-output neural network (MISONN) and multi-input multi-output neural network (MIMONN). A number of data sequences in varying sizes of different Anopheline malarial vectors and their corresponding species coding are employed to develop the neural network models. The classification efficiency of the network models for untrained data sequences is evaluated in terms of quantitative performance criteria. The results demonstrate the efficiency of the neural network models to extract the genetic information in ITS2 sequences and to adapt to new data. The method of MISONN is found to exhibit superior performance over MIMONN in distinguishing and identification of the mosquito vectors.     

Node-based differential network analysis in genomics

Gene dependency networks often undergo changes in response to different conditions. Understanding how these networks change across two conditions is an important task in genomics research. Most previous differential network analysis approaches assume that the difference between two condition-specific networks is driven by individual edges. Thus, they may fail in detecting key players which might represent important genes whose mutations drive the change of network. In this work, we develop a node-based differential network analysis (N-DNA) model to directly estimate the differential network that is driven by certain hub nodes. We model each condition-specific gene network as a precision matrix and the differential network as the difference between two precision matrices. Then we formulate a convex optimization problem to infer the differential network by combing a D-trace loss function and a row-column overlap norm penalty function. Simulation studies demonstrate that N-DNA provides more accurate estimate of the differential network than previous competing approaches. We apply N-DNA to ovarian cancer and breast cancer gene expression data. The model rediscovers known cancer-related genes and contains interesting predictions.     

Studies on Hydrogen Bonding Network Structures of Konjac Glucomannan

In this paper, the hydrogen bonding network models of konjac glucomannan （KGM） are predicted in the approach of molecular dynamics （MD）. These models have been proved by experiments whose results are consistent with those from simulation. The results show that the hydrogen bonding network structures of KGM are stable and the key linking points of hydrogen bonding network are at the O（6） and O（2） positions on KGM ring. Moreover, acety has significant influence on hydrogen bonding network and hydrogen bonding network structures are more stable after deacetylation.     

Double network hydrogels with controlled shape deformation: A mini review

Double network hydrogels (DN gels), consisting of two networks with strongly asymmetric network structures and properties, are one of most investigated high strength hydrogels. In most cases, the first network of DN gels is rigid, brittle and tightly crosslinked, while the second network is soft, ductile and loosely crosslinked. Because of the tunable and diverse network structures, DN gels with controlled shape deformation have attracted great attention in recent years. The shape deformation of DN gels can be controlled by first network, second network, or both networks. In this mini review, the shape deformation of DN gels via different networks will be summarized, and the application and future perspectives also are discussed. © 2018 Wiley Periodicals, Inc. J. Polym. Sci., Part B: Polym. Phys. 2018 , 56, 1351–1362     

粉末药品安体舒通的无损定量分析——人工神经网络－近红外光谱法的应用

研究了应用人工神经网络进行粉末药品的无损定量分析,使用安体舒通粉末药品的近红外漫反射光谱数据建立人工神经网络模型,预测未知样品,讨论了影响网络的各参数,使用了逼近度作为网络新的评价标准。     

Dielectric relaxation of liquid crystal molecules in anisotropic confinements

The dielectric relaxation of the liquid crystal 4-n-pentyl-4'-cyanobiphenyl (K15) in the presence of an anisotropic network has been studied. Anisotropic networks containing K15 molecules were prepared by in situ polymerisation of liquid-crystalline diacrylate molecules in a mixture containing K15. By changing the network concentration, the effect of the network molecules on the behaviour of the K15 molecules, which were not chemically attached to the network, was investigated. With increasing network concentration it was found that the mean relaxation times of K15 molecules shifted to lower temperatures and that their distribution became broader. The activation energy associated with the relaxation, however, remained almost constant before showing some increase at high network concentrations.     

Design of a tripartite network for the prediction of drug targets

Drug-target networks have aided in many target prediction studies aiming at drug repurposing or the analysis of side effects. Conventional drug-target networks are bipartite. They contain two different types of nodes representing drugs and targets, respectively, and edges indicating pairwise drug-target interactions. In this work, we introduce a tripartite network consisting of drugs, other bioactive compounds, and targets from different sources. On the basis of analog relationships captured in the network and so-called neighbor targets of drugs, new drug targets can be inferred. The tripartite network was found to have a stable structure and simulated network growth was accompanied by a steady increase in assortativity, reflecting increasing correlation between degrees of connected nodes leading to even network connectivity. Local drug environments in the tripartite network typically contained neighbor targets and revealed interesting drug-compound-target relationships for further analysis. Candidate targets were prioritized. The tripartite network design extends standard drug-target networks and provides additional opportunities for drug target prediction.     

Predicting coal ash fusion temperature based on its chemical composition using ACO-BP neural network

Coal ash fusion temperature is important to boiler designers and operators of power plants. Fusion temperature is determined by the chemical composition of coal ash, however, their relationships are not precisely known. A novel neural network, ACO-BP neural network, is used to model coal ash fusion temperature based on its chemical composition. Ant colony optimization (ACO) is an ecological system algorithm, which draws its inspiration from the foraging behavior of real ants. A three-layer network is designed with 10 hidden nodes. The oxide contents consist of the inputs of the network and the fusion temperature is the output. Data on 80 typical Chinese coal ash samples were used for training and testing. Results show that ACO-BP neural network can obtain better performance compared with empirical formulas and BP neural network. The well-trained neural network can be used as a useful tool to predict coal ash fusion temperature according to the oxide contents of the coal ash.     

软质聚氯乙烯分子网络及其Gaussian模量

采用应力松弛实验及Haward模型, 研究了增塑剂含量、填料[CaCO3、炭黑(CB)]和丁腈橡胶(NBR)对软质聚氯乙烯(PPVC)的分子链缠结网络结构、分子链滑移及Gaussian模量的影响. 结果表明, 在PPVC主网络达到极限伸长之前, PPVC材料的粘弹行为能很好地符合Haward模型. 增塑剂、CaCO3和CB虽然不改变主网络的缠结结构, 并且主网络的极限伸长不变, 但增塑剂可以降低主网络的网链密度, 而CaCO3和CB可以提高主网络的网链密度; 同时增塑剂可减弱次级网络, 增大PVC分子链滑移, 使材料的Gaussian模量下降; CaCO3和CB可增强次级网络, 减小PVC分子链滑移, 使材料的Gaussian模量增加. NBR的加入可以改变主网络的缠结结构, 增加主网络的极限伸长; 既可降低PPVC主网络的缠结密度, 又可减弱次级网络, 使Gaussian模量降低.     

Parallel Fock matrix construction program for molecular orbital calculation--specific computer with a hierarchical network

A parallel Fock matrix construction program for a hierarchical network has been developed on the molecular orbital calculation-specific EHPC system. To obtain high parallelization efficiency on the hierarchical network system, a multilevel dynamic load-balancing scheme was adopted, which provides equal load balance and localization of communications on a tree-structured hierarchical network. The parallelized Fock matrix construction routine was implemented into a GAMESS program on the EHPC system, which has a tree-structured hierarchical network. Benchmark results on a 63-processor system showed high parallelization efficiency even on the tree-structured hierarchical network.     

人工神经网络-分光光度法同时测定合金钢样中的铈组各个稀土元素

以二溴对甲基偶氮璜（DBM－SA）为显色剂,应用化学计量学中人工神经网络原理,结合分光光度法,对吸收光谱严重重叠的铈组五个元素不经分离可直接进行同时测定。对合金钢试样中铈组五个元素的个别含量及铈组总量的测定,结果令人满意。并进一步讨论了人工神经网络的结构及参数对分析结果的影响。     

A strategy based on protein-protein interface motifs may help in identifying drug off-targets

Networks are increasingly used to study the impact of drugs at the systems level. From the algorithmic standpoint, a drug can "attack" nodes or edges of a protein-protein interaction network. In this work, we propose a new network strategy, "The Interface Attack", based on protein-protein interfaces. Similar interface architectures can occur between unrelated proteins. Consequently, in principle, a drug that binds to one has a certain probability of binding to others. The interface attack strategy simultaneously removes from the network all interactions that consist of similar interface motifs. This strategy is inspired by network pharmacology and allows inferring potential off-targets. We introduce a network model that we call "Protein Interface and Interaction Network (P2IN)", which is the integration of protein-protein interface structures and protein interaction networks. This interface-based network organization clarifies which protein pairs have structurally similar interfaces and which proteins may compete to bind the same surface region. We built the P2IN with the p53 signaling network and performed network robustness analysis. We show that (1) "hitting" frequent interfaces (a set of edges distributed around the network) might be as destructive as eleminating high degree proteins (hub nodes), (2) frequent interfaces are not always topologically critical elements in the network, and (3) interface attack may reveal functional changes in the system better than the attack of single proteins. In the off-target detection case study, we found that drugs blocking the interface between CDK6 and CDKN2D may also affect the interaction between CDK4 and CDKN2D.