An enhanced protein secondary structure prediction using deep learning framework on hybrid profile based features

Accurate protein secondary structure prediction (PSSP) is essential to identify structural classes, protein folds, and its tertiary structure. To identify the secondary structure, experimental methods exhibit higher precision with the trade-off of high cost and time. In this study, we propose an effective prediction model which consists of hybrid features of 42-dimensions with the combination of convolutional neural network (CNN) and bidirectional recurrent neural network (BRNN). The proposed model is accessed on four benchmark datasets such as CB6133, CB513, CASP10, and CAP11 using Q3, Q8, and segment overlap (Sov) metrics. The proposed model reported Q3 accuracy of 85.4%, 85.4%, 83.7%, 81.5%, and Q8 accuracy 75.8%, 73.5%, 72.2%, and 70% on CB6133, CB513, CASP10, and CAP11 datasets respectively. The results of the proposed model are improved by a minimum factor of 2.5% and 2.1% in Q3 and Q8 accuracy respectively, as compared to the popular existing models on CB513 dataset. Further, the quality of the Q3 results is validated by structural class prediction and compared with PSI-PRED. The experiment showed that the quality of the Q3 results of the proposed model is higher than that of PSI-PRED.     

基于多重进化矩阵的蛋白质特征向量构造方法

特征向量的构造是蛋白质二级结构预测的一个关键问题. 现有的研究方法,通常只使用BLOSUM62进化矩阵生成PSSM矩阵,对蛋白质进化过程中存在的氨基酸残基突变现象缺乏考虑. 本文提出利用多重进化矩阵构造蛋白质特征向量,其融合了不同进化时间的PSSM矩阵,不仅能够很好地反映序列中氨基酸的位置信息,而且能够反映序列进化过程中氨基酸位点发生突变产生的影响. 本文通过组合不同进化程度的矩阵来构造特征向量,选用逻辑回归、随机森林和多分类支持向量机三种分类算法作为预测工具,利用网格搜索法和交叉实验法优化参数,在RS126、CB513和25PDB公用数据集上进行了若干组实验. 对比实验结果表明,本文所提出基于多重进化矩阵的蛋白质特征向量构造方法能够有效提高蛋白质二级结构的预测精度.     

应用ANN/HMM混合模型预测蛋白质二级结构

针对3状态隐马尔可夫模型（hidden Markov model,HMM）预测蛋白质二级结构准确率不高的问题,提出15状态HMM,通过改进的算法与BP神经网络相结合进行二级结构预测。研究对象为CB513数据集中筛选出的492条蛋白质序列,将其随机均分7组。应用混合模型进行预测,对准确率进行7交叉验证,Q3准确率达7721%,SOV值为7252%。结果表明,混合模型既能充分考虑相邻氨基酸残基间的相互影响,也能在一定程度上照顾二级结构的远程相关性,因此带来了较好的预测准确率。     

A new intelligent prediction system model-the compound pyramid model

A current development trend in research on intelligent systems is to optimize a general intelligent prediction system into an individuation intelligent prediction system that is applied in specialized fields. Protein structure prediction is a challenging international issue. In this paper, we propose a new intelligent prediction system model, designed as a multi-layer compound pyramid model, for predicting secondary protein structure. The model comprises four independent intelligent interfaces and several knowledge discovery methods. The model penetrates throughout the domain knowledge, with the effective attributes chosen by Causal Cellular Automata. Furthermore, a high pure structure database is constructed for training. On the RS126 dataset, the overall state per-residue accuracy, Q3, reached 83.99%, while on the CB513 dataset, Q3 reached 85.58%. Meanwhile, on the CASP8 sequences, the results are superior to those produced by other methods, such as Psipred, Jpred, APSSP2 and BehairPred. These results confirm that our method has a strong generalization ability, and that it provides a model for the construction of other intelligent systems.     

Predicting protein secondary structure using a mixed-modal SVM method in a compound pyramid model

Accurate protein secondary structure prediction plays an important role in direct tertiary structure modeling, and can also significantly improve sequence analysis and sequence-structure threading for structure and function determination. Hence improving the accuracy of secondary structure prediction is essential for future developments throughout the field of protein research.In this article, we propose a mixed-modal support vector machine (SVM) method for predicting protein secondary structure. Using the evolutionary information contained in the physicochemical properties of each amino acid and a position-specific scoring matrix generated by a PSI-BLAST multiple sequence alignment as input for a mixed-modal SVM, secondary structure can be predicted at significantly increased accuracy. Using a Knowledge Discovery Theory based on the Inner Cognitive Mechanism (KDTICM) method, we have proposed a compound pyramid model, which is composed of three layers of intelligent interface that integrate a mixed-modal SVM (MMS) module, a modified Knowledge Discovery in Databases (KDD1) process, a mixed-modal back propagation neural network (MMBP) module and so on.Testing against data sets of non-redundant protein sequences returned values for the Q₃ accuracy measure that ranged from 84.0% to 85.6%,while values for the SOV99 segment overlap measure ranged from 79.8% to 80.6%. When compared using a blind test dataset from the CASP8 meeting against currently available secondary structure prediction methods, our new approach shows superior accuracy.Availability: http://www.kdd.ustb.edu.cn/protein_Web/.     

A compact hybrid feature vector for an accurate secondary structure prediction

Amino acid propensity score is one of the earliest successful methods used in protein secondary structure prediction. However, the score performs poorly on small-sized datasets and low-identity protein sequences. Based on current in silico method, secondary structure can be predicted from local folds or local protein structure. In biology, the evolution of secondary structure produces local protein structure with different lengths. To precisely predict secondary structures, we propose a derivative feature vector, DPS that utilizes the optimal length of the local protein structure. DPS is the unification of amino acid propensity score and dihedral angle score. This new feature vector is further normalized to level the edges. Prediction is performed by support vector machines (SVM) over the DPS feature vectors with class labels generated by secondary structure assignment method (SSAM) and secondary structure prediction method (SSPM). All experiments are carried out on RS126 sequences. The results from this proposed method also highlight the overall accuracy of our method compared to other state-of-the-art methods. The performance of our method was acceptable specifically in dealing with low number and low identity sequences.     

Towards designing modular recurrent neural networks in learning protein secondary structures

Precise prediction of protein secondary structures from the associated amino acids sequence is of great importance in bioinformatics and yet a challenging task for machine learning algorithms. As a major step toward predicting the ultimate three dimensional structures, the secondary structure assignment specifies the protein function. Considering a multilayer perceptron neural network, pruned for optimum size of hidden layers, as the reference network, advanced kinds of recurrent neural network (RNN) are devised in this article to enhance the secondary structure prediction. To better model the strong correlations between secondary structure elements, types of modular reciprocal recurrent neural networks (MRR-NN) are examined. Additionally, to take into account the long-range interactions between amino acids in formation of the secondary structure, bidirectional RNN are investigated. A multilayer bidirectional recurrent neural network (MBR-NN) is finally applied to capture the predominant long-term dependencies. Eventually, a modular prediction system based on the interactive combination of the MRR-NN and MBR-NN boosts the percentage accuracy (Q₃) up to 76.91% and augments the segment overlap (SOV) up to 68.13% when tested on the PSIPRED dataset. The coupling effects of the secondary structure types as well as the sequential information of amino acids along the protein chain can be well cast by the integration of the MRR-NN and the MBR-NN.     

蛋白质二级结构预测方法的评价

蛋白质结构预测是后基因组时代的一项重要任务，蛋白质二级结构预测是蛋白质结构预测的关键步骤。现在一般认为，如果蛋白质二级结构的预测准确率达到80％的话，就可以基本准确地预测一个蛋白质分子的三维空间结构。目前蛋白质二级结构预测的方法不断涌现，提供二级结构预测的网站也逐渐增多。为给广大研究工作者在选择使用这些预测方法时提供一种参考，文章采用统一的标准对10种比较重要而且有效的方法进行测试，并在此基础上做出评价和分析，这10种方法是：GORI、PROF、GORⅣ、NNPREDICT、PHDsec、SSpro v 2．0、PSIPRED、PREDATOR、SOPMA和APSSP2。比较结果显示：APSSP2、SSpro v 2．0和PSIPRED方法的预测效果较好，可以作为使用时的首选方案，其中尤其以APSSP2方法的预测效果最佳。     

蛋白质二级结构预测的一种新的编码方式

编码方式是影响蛋白质二级结构预测准确率的重要因素之一。针对单序列蛋白质二级结构预测问题,提出了一种新的综合编码方法。该编码是根据氨基酸出现在每种二级结构中的倾向因子以及氨基酸的疏水性值进行分类,并以二进制形式来表示每类氨基酸的编码方法。在相同的实验条件下,首先用不同的编码方式对数据集CB513进行编码,然后采用支持向量机的方法进行训练建模预测。实验结果显示提出编码的预测准确率比20位正交编码和5位编码分别高出1.48%和10.68%。可见,该编码比较适合非同源或低同源蛋白质结构预测。     

基于条件随机场进行蛋白质二级结构预测*

使用一种新的概率图模型——条件随机场对蛋白质二级结构进行预测,并给出了模型的构建、训练以及解码的算法。应用这一模型对一个典型的蛋白质数据集CB513的二级结构进行了预测,并将预测结果与其他方法进行比较,预测准确度有明显的提高。     

基于深度学习的八类蛋白质二级结构预测算法

蛋白质二级结构预测是结构生物学中的一个重要问题。针对八类蛋白质二级结构预测,提出了一种基于递归神经网络和前馈神经网络的深度学习预测算法。该算法通过双向递归神经网络建模氨基酸间的局部和长程相互作用,递归神经网络的隐层输出进一步送入到三层的前馈神经网络以便进行八类蛋白质二级结构预测。实验结果表明,提出的算法在CB513数据集上达到了67.9%的Q₈预测精度,显著地优于SSpro8和SC-GSN。     

Improving protein secondary structure predictions by prediction fusion

Protein secondary structure prediction is still a challenging problem at today. Even if a number of prediction methods have been presented in the literature, the various prediction tools that are available on-line produce results whose quality is not always fully satisfactory. Therefore, a user has to know which predictor to use for a given protein to be analyzed. In this paper, we propose a server implementing a method to improve the accuracy in protein secondary structure prediction. The method is based on integrating the prediction results computed by some available on-line prediction tools to obtain a combined prediction of higher quality. Given an input protein p whose secondary structure has to be predicted, and a group of proteins F, whose secondary structures are known, the server currently works according to a two phase approach: (i) it selects a set of predictors good at predicting the secondary structure of proteins in F (and, therefore, supposedly, that of p as well), and (ii) it integrates the prediction results delivered for p by the selected team of prediction tools. Therefore, by exploiting our system, the user is relieved of the burden of selecting the most appropriate predictor for the given input protein being, at the same time, assumed that a prediction result at least as good as the best available one will be delivered. The correctness of the resulting prediction is measured referring to EVA accuracy parameters used in several editions of CASP.     

An optimized ANN model based on genetic algorithm for predicting ripping production

Due to the environmental constraints and the limitations on blasting, ripping as a ground loosening and breaking method has become more popular in both mining and civil engineering applications. As a result, a more applicable rippability model is required to predict ripping production (Q) before conducting such tests. In this research, a hybrid genetic algorithm (GA) optimized by artificial neural network (ANN) was developed to predict ripping production results obtained from three sites in Johor state, Malaysia. It should be noted that the mentioned hybrid model was first time applied in this field. In this regard, 74 ripping tests were investigated in the studied areas and the relevant parameters were also measured. A series of GA–ANN models were conducted in order to propose a hybrid model with a higher accuracy level. To demonstrate the performance capacity of the hybrid GA–ANN model, a pre-developed ANN model was also proposed and results of predictive models were compared using several performance indices. The results revealed higher accuracy of the proposed hybrid GA–ANN model in estimating Q compared to ANN technique. As an example, root-mean-square error values of 0.092 and 0.131 for testing datasets of GA–ANN and ANN techniques, respectively, express the superiority of the newly developed model in predicting ripping production.

     

Global and Local (Glocal) Bagging Approach for Classifying Noisy Dataset

Learning from noisy data is a challenging task for data mining research. In this paper, we argue that for noisy data both global bagging strategy and local bagging strategy su er from their own inherent disadvantages and thus cannot form accurate prediction models. Consequently, we present a Global and Local Bagging (called Glocal Bagging:GB) approach to tackle this problem. GB assigns weight values to the base classi ers under the consideration that: (1) for each test instance I_x, GB prefers bags close to I_x, which is the nature of the local learning strategy; (2) for base classi ers, GB assigns larger weight values to the ones with higher accuracy on the out-of-bag, which is the nature of the global learning strategy. Combining (1) and (2), GB assign large weight values to the classi ers which are close to the current test instance I_x and have high out-of-bag accuracy. The diversity/accuracy analysis on synthetic datasets shows that GB improves the classi er ensemble's performance by increasing its base classi er's accuracy. Moreover, the bias/variance analysis also shows that GB's accuracy improvement mainly comes from the reduction of the bias error. Experiment results on 25 UCI benchmark datasets show that when the datasets are noisy, GB is superior to other former proposed bagging methods such as the classical bagging, bragging, nice bagging, trimmed bagging and lazy bagging.     

Rotationally symmetric FDTD for wideband performance prediction of TM01 DR filters

The generalized perfectly matched layer (GPML) coupled with rotationally symmetric (RS)‐FDTD method has been utilized to extract the S‐parameters for several probe‐coupled TM₀₁ dielectric resonator (DR) filters to directly obtain the theoretical wideband spurious performance. The computationally efficient (RS)‐FDTD method has also been used to obtain accurate filter parameters for TE₀₁ and TM₀₁ dielectric resonators loaded in cylindrical cavities. The RS‐FDTD method combined with digital filtering and the Matrix Pencil technique are used to analyze the resonant frequencies, inter‐resonator coupling, and external Q values. When perturbation theory is used with RS‐FDTD, accurate values of unloaded Q are obtained. © 2002 Wiley Periodicals, Inc. Int J RF and Microwave CAE 12: 259–271, 2002.     

A multimodal neural network with single-state predictions for protein secondary structure

Prediction of protein secondary structure is considered to be an important step toward elucidating the three-dimensional structure and function of proteins. We have developed a multimodal neural network (MNN) to predict protein secondary structure. The MNN is composed of several subclassifiers for single-state predictions using neural networks and a decision neural network (DNN). Each subclassifier employs a number of subnetworks to predict the single-state of the secondary structure individually and produces the final results by majority decision. The DNN uses a three-layer neural network to produce the final overall prediction from the outputs of the single-state predictions. The MNN gives an overall accuracy of 71.1% with corresponding Matthews correlation coefficients of C_H = 0.62 and C_E = 0.53. The prediction test is based on a database of 126 nonhomologous protein sequences. This work was presented, in part, at the 8th International Symposium on Artificial Life and Robotics, Oita, Japan, January 24#x2013;26, 2003.     

Two-level and hybrid ensembles of decision trees for high performance concrete compressive strength prediction

Accurate prediction of high performance concrete (HPC) compressive strength is very important issue. In the last decade, a variety of modeling approaches have been developed and applied to predict HPC compressive strength from a wide range of variables, with varying success. The selection, application and comparison of decent modeling methods remain therefore a crucial task, subject to ongoing researches and debates. This study proposes three different ensemble approaches: (i) single ensembles of decision trees (DT) (ii) two-level ensemble approach which employs same ensemble learning method twice in building ensemble models (iii) hybrid ensemble approach which is an integration of attribute-base ensemble method (random sub-spaces RS) and instance-base ensemble methods (bagging Bag, stochastic gradient boosting GB). A decision tree is used as the base learner of ensembles and its results are benchmarked to proposed ensemble models. The obtained results show that the proposed ensemble models could noticeably advance the prediction accuracy of the single DT model and for determining average determination of correlation, the best models for HPC compressive strength forecasting are GB–RS DT, RS–GB DT and GB–GB DT among the eleven proposed predictive models, respectively. The obtained results show that the proposed ensemble models could noticeably advance the prediction accuracy of the single DT model and for determining determination of correlation (R²_max), the best models for HPC compressive strength forecasting are GB–RS DT (R²=0.9520), GB–GB DT (R²=0.9456) and Bag–Bag DT (R²=0.9368) among the eleven proposed predictive models, respectively.     

一种采用新型聚类方法的最佳类簇数确定算法

聚类分析是统计学、模式识别和机器学习等领域的研究热点.通过有效的聚类分析,数据集的内在结构与特征可以被很好地发掘出来.然而,无监督学习的特性使得当前已有的聚类方法依旧面临着聚类效果不稳定、无法对多种结构的数据集进行正确聚类等问题.针对这些问题,首先将K-means算法和层次聚类算法的聚类思想相结合,提出了一种混合聚类算...     

HU-FCF: A hybrid user-based fuzzy collaborative filtering method in Recommender Systems

Recommender Systems (RS) have been being captured a great attraction of researchers by their applications in various interdisciplinary fields. Fuzzy Recommender Systems (FRS) is an extension of RS with the fuzzy similarity being calculated based on the users’ demographic data instead of the hard user-based degree. Based upon the observations that the FRS researches did not offer a mathematical definition of FRS accompanied with its algebraic operations and properties, and the fuzzy similarity degree is not enough to express accurately the analogousness between users, in this paper we will present a systematic mathematical definition of FRS including theoretical analyses of algebraic operations and properties and propose a novel hybrid user-based fuzzy collaborative filtering method that integrates the fuzzy similarity degrees between users based on the demographic data with the hard user-based degrees calculated from the rating histories into the final similarity degrees in order to obtain high accuracy of prediction. Experimental results on some benchmark datasets show that the proposed method obtains better accuracy than other relevant methods. Lastly, an application for the football results prediction is given to illustrate the uses of the proposed method.     

正交约束多头自注意力的场景文本识别

目的 场景文本识别（scene text recognition，STR）是计算机视觉中的一个热门研究领域。最近，基于多头自注意力机制的视觉Transformer （vision Transformer，ViT）模型被提出用于STR，以实现精度、速度和计算负载的平衡。然而，没有机制可以保证不同的自注意力头确实捕捉到多样性的特征，这将导致使用多头自注意力机制的ViT模型在多样性极强的场景文本识别任务中表现不佳。针对这个问题，提出了一种新颖的正交约束来显式增强多个自注意力头之间的多样性，提高多头自注意力对不同子空间信息的捕获能力，在保证速度和计算效率的同时进一步提高网络的精度。方法 首先提出了针对不同自注意力头上Q （query）、K （key）和V （value）特征的正交约束，这可以使不同的自注意力头能够关注到不同的查询子空间、键子空间、值子空间的特征，关注不同子空间的特征可以显式地使不同的自注意力头捕捉到更具差异的特征。还提出了针对不同自注意力头上Q 、K 和V 特征线性变换权重的正交约束，这将为Q、K和V特征的学习提供正交权重空间的解决方案，并在网络训练中带来隐式正则化的效果。结果 实验在7个数据集上与基准方法进行比较，在规则数据集Street View Text （SVT）上精度提高了0.5%；在不规则数据集CUTE80 （CT）上精度提高了1.1%；在7个公共数据集上的整体精度提升了0.5%。结论 提出的即插即用的正交约束能够提高多头自注意力机制在STR任务中的特征捕获能力，使ViT模型在STR任务上的识别精度得到提高。本文代码已公开： https：//github.com/lexiaoyuan/XViTSTR。