A neural network for predicting the stability of DNA/DNA duplexes

A back-propagation neural network method has been developed to predict the stability of DNA/DNA duplexes. Calculated Tm with the present parameters fits the experimental values within reasonable errors (AD = 1.86 K, SEP = 1.99151, R2 = 0.9894 for NN1; AD = 1.59667 K, SEP = 2.03824, R2 = 0.99371 for NN2) and it has the advantage that the determinations of thermodynamic parameters are not needed.     

Phosphorylation of tropomyosin extends cooperative binding of myosin beyond a single regulatory unit

Tropomyosin (Tm) is one of the major phosphoproteins comprising the thin filament of muscle. However, the specific role of Tm phosphorylation in modulating the mechanics of actomyosin interaction has not been determined. Here we show that Tm phosphorylation is necessary for long-range cooperative activation of myosin binding. We used a novel optical trapping assay to measure the isometric stall force of an ensemble of myosin molecules moving actin filaments reconstituted with either natively phosphorylated or dephosphorylated Tm. The data show that the thin filament is cooperatively activated by myosin across regulatory units when Tm is phosphorylated. When Tm is dephosphorylated, this "long-range" cooperative activation is lost and the filament behaves identically to bare actin filaments. However, these effects are not due to dissociation of dephosphorylated Tm from the reconstituted thin filament. The data suggest that end-to-end interactions of adjacent Tm molecules are strengthened when Tm is phosphorylated, and that phosphorylation is thus essential for long range cooperative activation along the thin filament.     

Short-term prediction of chaotic time series by using RBF network with regression weights

We propose a framework for constructing and training a radial basis function (RBF) neural network. The structure of the gaussian functions is modified using a pseudo-gaussian function (PG) in which two scaling parameters sigma are introduced, which eliminates the symmetry restriction and provides the neurons in the hidden layer with greater flexibility with respect to function approximation. We propose a modified PG-BF (pseudo-gaussian basis function) network in which the regression weights are used to replace the constant weights in the output layer. For this purpose, a sequential learning algorithm is presented to adapt the structure of the network, in which it is possible to create a new hidden unit and also to detect and remove inactive units. A salient feature of the network systems is that the method used for calculating the overall output is the weighted average of the output associated with each receptive field. The superior performance of the proposed PG-BF system over the standard RBF are illustrated using the problem of short-term prediction of chaotic time series.     

Sorting of a nonmuscle tropomyosin to a novel cytoskeletal compartment in skeletal muscle results in muscular dystrophy

Tropomyosin (Tm) is a key component of the actin cytoskeleton and >40 isoforms have been described in mammals. In addition to the isoforms in the sarcomere, we now report the existence of two nonsarcomeric (NS) isoforms in skeletal muscle. These isoforms are excluded from the thin filament of the sarcomere and are localized to a novel Z-line adjacent structure. Immunostained cross sections indicate that one Tm defines a Z-line adjacent structure common to all myofibers, whereas the second Tm defines a spatially distinct structure unique to muscles that undergo chronic or repetitive contractions. When a Tm (Tm3) that is normally absent from muscle was expressed in mice it became associated with the Z-line adjacent structure. These mice display a muscular dystrophy and ragged-red fiber phenotype, suggestive of disruption of the membrane-associated cytoskeletal network. Our findings raise the possibility that mutations in these tropomyosin and these structures may underpin these types of myopathies.     

Dynamics of the muscle thin filament regulatory switch: the size of the cooperative unit.

Actin thin filaments containing bound tropomyosin (Tm) or tropomyosin troponin (Tm.Tn) exist in two states ("off" and "on") with different affinities for myosin heads (S1), which results in the cooperative binding of S1. The rate of S1 binding to, and dissociating from, actin, Tm.actin, and Tm.Tn.actin, monitored by light scattering (LS), was compared with the rate of change in state, monitored by the excimer fluorescence (Fl) of a pyrene label attached to Tm. The ATP-induced S1 dissociation showed similar exponential decreases in LS for actin.S1, Tm.actin.S1, and Tm.Tn.actin.S1 +/- Ca2+. The Fl change, however, showed a delay that was greater for Tm.Tn.actin than Tm.actin, independent of Ca2+. The S1 binding kinetics gave observed rate constants for the S1-induced change in state that were 5-6 times the observed rate constants of S1 binding to Tm.actin, which were increased to 10-12 for Tm.Tn.actin, independent of Ca2+. The rate of the Fl signals showed that the on/off states were in rapid equilibrium. These data indicate that the apparent cooperative unit for Tm.actin is 5-6 actin subunits rather than the minimum structural unit size of 7, and is increased to 10-12 subunits for Tm.Tn.actin, independent of the presence of Ca2+. Thus, Tm appears semi-flexible, and Tn increases communication between neighboring structural units. A general model for the dynamic transitions involved in muscle regulation is presented.     

Feedforward neural network with adaptive reference pattern layer.

A hybrid neural network architecture is investigated for modeling purposes. The proposed hybrid is based on the multilayer perceptron (MLP) network. In addition to the usual hidden layers, the first hidden layer is selected to be an adaptive reference pattern layer. Each unit in this new layer incorporates a reference pattern that is located somewhere in the space spanned by the input variables. The outputs of these units are the component wise-squared differences between the elements of a reference pattern and the inputs. The reference pattern layer has some resemblance to the hidden layer of the radial basis function (RBF) networks. Therefore the proposed design can be regarded as a sort of hybrid of MLP and RBF networks. The presented benchmark experiments show that the proposed hybrid can provide significant advantages over standard MLPs and RBFs in terms of fast and efficient learning, and compact network structure.     

Thermal stability determinants of chicken egg-white lysozyme core mutants: hydrophobicity, packing volume, and conserved buried water molecules.

A series of 24 mutants was made in the buried core of chicken lysozyme at positions 40, 55, and 91. The midpoint temperature of thermal denaturation transition (Tm) values of these core constructs range from 60.9 to 77.3 degrees C, extending an earlier, more limited investigation on thermostability. The Tm values of variants containing conservative replacements for the wild type (WT) (Thr 40-Ile 55-Ser 91) triplet are linearly correlated with hydrophobicity (r = 0.81) and, to a lesser degree, with combined side-chain volume (r = 0.75). The X-ray structures of the S91A (1.9 A) and I55L/S91T/D101S (1.7 A) mutants are presented. The former amino acid change is found in duck and mammalian lysozymes, and the latter contains the most thermostable core triplet. A network of four conserved, buried water molecules is associated with the core. It is postulated that these water molecules significantly influence the mutational tolerance at the individual triplet positions. The pH dependence of Tm for the S91D mutant was compared with that of WT enzyme. The pKa of S91D is 1.2 units higher in the native than in the denatured state, corresponding to delta delta G298 = 1.7 kcal/mol. This is a low value for charge burial and likely reflects the moderating influence of the buried water molecules or a conformational change. Thermal and chemical denaturation and far UV CD spectroscopy were used to characterize the in vitro properties of I55T. This variant, which buries a hydroxyl group, has similar properties to those of the human amyloidogenic variant I56T.     

Application of the Free Energy Calculations to Study Drug-enzyme and Drug-dna Complexes

Applications of two free energy calculation approaches are presented to study drug-biomolecule complexes. The first method, the free energy perturbation (FEP) method and molecular dynamics simulations has been applied to study the JG-365 inhibitor bound to the HIV-aspartic protease. The FEP method has been applied to predict the consequence of replacing each of the seven peptide bonds in the JG-365 by trans-ethylene or fluoroethylene units. The necessary initial conformations of the inhibitor for "in water" perturbations have been found using neural network clustering approach applied to the long molecular dynamics trajectory of the inhibitor in water solution. The second method is applied to study binding free energies of some DNA-drug complexes and is based on analysis of long molecular dynamics trajectories by continuum solvent approach (MM/PBSA).     

Phase transition in dioctadecyldimethylammonium bromide and chloride vesicles prepared by different methods

The gel to liquid crystalline phase transition of the double-chained cationic dioctadecyldimethylammonium chloride and bromide (DODAX, X = Cl- or Br-) in aqueous vesicle dispersions prepared by non-sonication. sonication and extrusion has been investigated using high-sensitivity differential scanning calorimetry (DSC). The transition temperature (Tm) is a function of the preparation method, amphiphile concentration, vesicle curvature and nature of the counterion. DSC thermograms for DODAB and DODAC non-sonicated vesicle dispersions exhibit a single endothermic peak at Tm roughly independent of concentration up to 10 mM. Extrusion broadens the transition peak and shifts Tm downwards. Sonication, however, broadens slightly the transition peak and tends to shift Tm upwards suggesting that extrusion and sonication form vesicles with different characteristics. DODAC always exhibits higher Tm than DODAB irrespective of the preparation method. Tm changes as follows: Tm (sonicated) > or = Tm (non-sonicated) > Tm (extruded). Hysteresis of about 7 degrees C was observed for DODAB vesicle dispersions.     

Persistence length of human cardiac α-tropomyosin measured by single molecule direct probe microscopy

α-Tropomyosin (αTm) is the predominant tropomyosin isoform in adult human heart and constitutes a major component in Ca2?-regulated systolic contraction of cardiac muscle. We present here the first direct probe images of WT human cardiac αTm by atomic force microscopy, and quantify its mechanical flexibility with three independent analysis methods. Single molecules of bacterially-expressed human cardiac αTm were imaged on poly-lysine coated mica and their contours were analyzed. Analysis of tangent-angle (θ(s)) correlation along molecular contours, second moment of tangent angles (<θ2(s)>), and end-to-end length (L(e-e)) distributions respectively yielded values of persistence length (L(p)) of 41-46 nm, 40-45 nm, and 42-52 nm, corresponding to 1-1.3 molecular contour lengths (L(c)). We also demonstrate that a sufficiently large population, with at least 100 molecules, is required for a reliable L(p) measurement of αTm in single molecule studies. Our estimate that L(p) for αTm is only slightly longer than L(c) is consistent with a previous study showing there is little spread of cooperative activation into near-neighbor regulatory units of cardiac thin filaments. The L(p) determined here for human cardiac αTm perhaps represents an evolutionarily tuned optimum between Ca2? sensitivity and cooperativity in cardiac thin filaments and likely constitutes an essential parameter for normal function in the human heart.     

Solid phase synthesis of oligodeoxyribonucleoside phosphorodithioates from thiophosphoramidites.

Oligonucleoside phosphorodithioates 1 are modified DNA sequences with potential use as antisense oligonucleotides. The preparation of up to 20-mers containing all four bases by solid phase synthesis is described, with details on the preparation of the four monomer units (protected nucleoside thiophosphoramidites 2), the conditions used for the assembly of the strands with up to 19 phosphorodithioate linkages, and the purification and characterisation of the products. Full-length homogeneity of HPLC-purified all-phosphorodithioate products is demonstrated by PAGE, but 31P NMR discloses the presence of phosphorothioate impurities (typically 8-9%), the origin of which is discussed. Oligonucleoside phosphorodithioates are freely soluble in water at neutral or basic pH, and are very stable towards oxidation, hydrolysis, and nuclease cleavage. Their ability to hybridize to complementary DNA has been studied by UV melting point (Tm) measurements. The observed depression of Tm, 0.5-2 degrees C per phosphorodithioate linkage, is higher that the 0.4-0.6 degrees C found for phosphorothioates.     

A novel member of the YchN-like fold: solution structure of the hypothetical protein Tm0979 from Thermotoga maritima

We report herein the NMR structure of Tm0979, a structural proteomics target from Thermotoga maritima. The Tm0979 fold consists of four beta/alpha units, which form a central parallel beta-sheet with strand order 1234. The first three helices pack toward one face of the sheet and the fourth helix packs against the other face. The protein forms a dimer by adjacent parallel packing of the fourth helices sandwiched between the two beta-sheets. This fold is very interesting from several points of view. First, it represents the first structure determination for the DsrH family of conserved hypothetical proteins, which are involved in oxidation of intracellular sulfur but have no defined molecular function. Based on structure and sequence analysis, possible functions are discussed. Second, the fold of Tm0979 most closely resembles YchN-like folds; however the proteins that adopt these folds differ in secondary structural elements and quaternary structure. Comparison of these proteins provides insight into possible mechanisms of evolution of quaternary structure through a simple mechanism of hydrophobicity-changing mutations of one or two residues. Third, the Tm0979 fold is found to be similar to flavodoxin-like folds and beta/alpha barrel proteins, and may provide a link between these very abundant folds and putative ancestral half-barrel proteins.     

应用人工神经网络评价湖泊的富营养化

应用人工神经网络方法,以化学需氧量、总氮、总磷和透明度作为评价参数,经反复尝试,构建了具有４层结构用于评价湖泊富营养化的误差逆传播网络．其输入层有４个神经元,２个隐含层也各有４个神经元,输出层有１个神经元．以太湖富营养化评价标准作为样本模式提供给网络,按照误差逆传播网络的学习规则对网络进行训练,经过３７６８４次学习后,网络达到预先给定的收敛标准．使网络具备了识别湖泊富营养化程度的功能．应用该网络对我国１７个湖泊的富营养化程度进行评价,操作过程简便易行,评价结果切合实际,展示了这种方法的一系列优点．     

Prediction of continuous B-cell epitopes in an antigen using recurrent neural network

B-cell epitopes play a vital role in the development of peptide vaccines, in diagnosis of diseases, and also for allergy research. Experimental methods used for characterizing epitopes are time consuming and demand large resources. The availability of epitope prediction method(s) can rapidly aid experimenters in simplifying this problem. The standard feed-forward (FNN) and recurrent neural network (RNN) have been used in this study for predicting B-cell epitopes in an antigenic sequence. The networks have been trained and tested on a clean data set, which consists of 700 non-redundant B-cell epitopes obtained from Bcipep database and equal number of non-epitopes obtained randomly from Swiss-Prot database. The networks have been trained and tested at different input window length and hidden units. Maximum accuracy has been obtained using recurrent neural network (Jordan network) with a single hidden layer of 35 hidden units for window length of 16. The final network yields an overall prediction accuracy of 65.93% when tested by fivefold cross-validation. The corresponding sensitivity, specificity, and positive prediction values are 67.14, 64.71, and 65.61%, respectively. It has been observed that RNN (JE) was more successful than FNN in the prediction of B-cell epitopes. The length of the peptide is also important in the prediction of B-cell epitopes from antigenic sequences. The webserver ABCpred is freely available at www.imtech.res.in/raghava/abcpred/.     

A Study on Protein Residue Contacts Prediction by Recurrent Neural Network

1IntroductionThe three-dimensional(3D)structure of a proteinis perhaps the most important of all its features,since itdetermines completely how the protein functions andinteracts with other molecules.Most biological mech-anisms at the protein level are based on shape-complementarity,so that proteins present particularconcavities and convexities that allow them to bind toeach other and formcomplexstructures,and tendon.Forthis reason,for instance,the drug design problem con-sists primarily in th…     

Tropomyosin variants describe distinct functional subcellular domains in differentiated vascular smooth muscle cells

Tropomyosin (Tm) is known to be an important gatekeeper of actin function. Tm isoforms are encoded by four genes, and each gene produces several variants by alternative splicing, which have been proposed to play roles in motility, proliferation, and apoptosis. Smooth muscle studies have focused on gizzard smooth muscle, where a heterodimer of Tm from the α-gene (Tmsm-α) and from the β-gene (Tmsm-β) is associated with contractile filaments. In this study we examined Tm in differentiated mammalian vascular smooth muscle (dVSM). Liquid chromatography-tandem mass spectrometry (LC MS/MS) analysis and Western blot screening with variant-specific antibodies revealed that at least five different Tm proteins are expressed in this tissue: Tm6 (Tmsm-α) and Tm2 from the α-gene, Tm1 (Tmsm-β) from the β-gene, Tm5NM1 from the γ-gene, and Tm4 from the δ-gene. Tm6 is by far most abundant in dVSM followed by Tm1, Tm2, Tm5NM1, and Tm4. Coimmunoprecipitation and coimmunofluorescence studies demonstrate that Tm1 and Tm6 coassociate with different actin isoforms and display different intracellular localizations. Using an antibody specific for cytoplasmic γ-actin, we report here the presence of a γ-actin cortical cytoskeleton in dVSM cells. Tm1 colocalizes with cortical cytoplasmic γ-actin and coprecipitates with γ-actin. Tm6, on the other hand, is located on contractile bundles. These data indicate that Tm1 and Tm6 do not form a classical heterodimer in dVSM but rather describe different functional cellular compartments.     

Sympatry inference and network analysis in biogeography

A new approach for biogeography to find patterns of sympatry, based on network analysis, is proposed. Biogeographic analysis focuses basically on sympatry patterns of species. Sympatry is a network (= relational) datum, but it has never been analyzed before using relational tools such as Network Analysis. Our approach to biogeographic analysis consists of two parts: first the sympatry inference and second the network analysis method (NAM). The sympatry inference method was designed to propose sympatry hypothesis, constructing a basal sympatry network based on punctual data, independent of a priori distributional area determination. In this way, two or more species are considered sympatric when there is interpenetration and relative proximity among their records of occurrence. In nature, groups of species presenting within-group sympatry and between-group allopatry constitute natural units (units of co-occurrence). These allopatric units are usually connected by intermediary species. The network analysis method (NAM) that we propose here is based on the identification and removal of intermediary species to segregate units of co-occurrence, using the betweenness measure and the clustering coefficient. The species ranges of the units of co-occurrence obtained are transferred to a map, being considered as candidates to areas of endemism. The new approach was implemented on three different real complex data sets (one of them a classic example previously used in biogeography) resulting in (1) independence of predefined spatial units; (2) definition of co-occurrence patterns from the sympatry network structure, not from species range similarities; (3) higher stability in results despite scale changes; (4) identification of candidates to areas of endemism supported by strictly endemic species; (5) identification of intermediary species with particular biological attributes.     

Neural Network Cascade Optimizes MicroRNA Biomarker Selection for Nasopharyngeal Cancer Prognosis

MicroRNAs (miRNAs) have been shown to be promising biomarkers in predicting cancer prognosis. However, inappropriate or poorly optimized processing and modeling of miRNA expression data can negatively affect prediction performance. Here, we propose a holistic solution for miRNA biomarker selection and prediction model building. This work introduces the use of a neural network cascade, a cascaded constitution of small artificial neural network units, for evaluating miRNA expression and patient outcome. A miRNA microarray dataset of nasopharyngeal carcinoma was retrieved from Gene Expression Omnibus to illustrate the methodology. Results indicated a nonlinear relationship between miRNA expression and patient death risk, implying that direct comparison of expression values is inappropriate. However, this method performs transformation of miRNA expression values into a miRNA score, which linearly measures death risk. Spearman correlation was calculated between miRNA scores and survival status for each miRNA. Finally, a nine-miRNA signature was optimized to predict death risk after nasopharyngeal carcinoma by establishing a neural network cascade consisting of 13 artificial neural network units. Area under the ROC was 0.951 for the internal validation set and had a prediction accuracy of 83% for the external validation set. In particular, the established neural network cascade was found to have strong immunity against noise interference that disturbs miRNA expression values. This study provides an efficient and easy-to-use method that aims to maximize clinical application of miRNAs in prognostic risk assessment of patients with cancer.     

Quantitative measurement of two-component pH-sensitive colorimetric spectra using multilayer neural networks

The purpose of this research was to develop a noise tolerant and faster processing approach for in vivo and in vitro spectrophotometric applications where distorted spectra are difficult to interpret quantitatively. A PC based multilayer neural network with a sigmoid activation function and a generalized delta learning rule was trained with a two component (protonated and unprotonated form) pH-dependent spectrum generated from microspectrophotometry of the vital dye neutral red (NR). The network makes use of the digitized absorption spectrum between 375 and 675 nm. The number of nodes in the input layer was determined by the required resolution. The number of output nodes determined the step size of the quantization value used to distinguish the input spectra (i.e. defined the number of distinct output steps). Mathematic analysis provided the conditions for which this network is guaranteed to converge. Simulation results showed that features of the input spectrum were successfully identified and stored in the weight matrix of the input and hidden layers. After convergent training with typical spectra, a calibration curve was constructed to interpret the output layer activity and therefore, predict interpolated pH values of unknown spectra. With its built-in redundant presentation, this approach needed no preprocessing procedures (baseline correction or intensive signal averaging) normally used in multicomponent analyses. The identification of unknown spectra with the activities of the output layer is a one step process using the convergent weight matrix. After learning from examples, real time applications can be accomplished without solving multiple linear equations as in the multiple linear regression method. This method can be generalized to pattern oriented sensory information processing and multi-sensor data fusion for quantitative measurement purposes.     

Topology representing network enables highly accurate classification of protein images taken by cryo electron-microscope without masking

In single-particle analysis, a three-dimensional (3-D) structure of a protein is constructed using electron microscopy (EM). As these images are very noisy in general, the primary process of this 3-D reconstruction is the classification of images according to their Euler angles, the images in each classified group then being averaged to reduce the noise level. In our newly developed strategy of classification, we introduce a topology representing network (TRN) method. It is a modified method of a growing neural gas network (GNG). In this system, a network structure is automatically determined in response to the images input through a growing process. After learning without a masking procedure, the GNG creates clear averages of the inputs as unit coordinates in multi-dimensional space, which are then utilized for classification. In the process, connections are automatically created between highly related units and their positions are shifted where the inputs are distributed in multi-dimensional space. Consequently, several separated groups of connected units are formed. Although the interrelationship of units in this space are not easily understood, we succeeded in solving this problem by converting the unit positions into two-dimensional (2-D) space, and by further optimizing the unit positions with the simulated annealing (SA) method. In the optimized 2-D map, visualization of the connections of units provided rich information about clustering. As demonstrated here, this method is clearly superior to both the multi-variate statistical analysis (MSA) and the self-organizing map (SOM) as a classification method and provides a first reliable classification method which can be used without masking for very noisy images.