A molecular dynamics and knowledge-based computational strategy to predict native-like structures of polypeptides

One of the main research problems in structural bioinformatics is the prediction of three-dimensional structures (3-D) of polypeptides or proteins. The current rate at which amino acid sequences are identified increases much faster than the 3-D protein structure determination by experimental methods, such as X-ray diffraction and NMR techniques. The determination of protein structures is both experimentally expensive and time consuming. Predicting the correct 3-D structure of a protein molecule is an intricate and arduous task. The protein structure prediction (PSP) problem is, in computational complexity theory, an NP-complete problem. In order to reduce computing time, current efforts have targeted hybridizations between ab initio and knowledge-based methods aiming at efficient prediction of the correct structure of polypeptides. In this article we present a hybrid method for the 3-D protein structure prediction problem. An artificial neural network knowledge-based method that predicts approximated 3-D protein structures is combined with an ab initio strategy. Molecular dynamics (MD) simulation is used to the refinement of the approximated 3-D protein structures. In the refinement step, global interactions between each pair of atoms in the molecule (including non-bond interactions) are evaluated. The developed MD protocol enables us to correct polypeptide torsion angles deviation from the predicted structures and improve their stereo-chemical quality. The obtained results shows that the time to predict native-like 3-D structures is considerably reduced. We test our computational strategy with four mini proteins whose sizes vary from 19 to 34 amino acid residues. The structures obtained at the end of 32.0 nanoseconds (ns) of MD simulation were comparable topologically to their correspondent experimental structures.     

Prediction of protein structural class for low-similarity sequences using Chou’s pseudo amino acid composition and wavelet denoising

Prediction of protein structural class plays an important role in protein structure and function analysis, drug design and many other biological applications. Prediction of protein structural class for low-similarity sequences is still a challenging task. Based on the theory of wavelet denoising, this paper presents a novel method of prediction of protein structural class for the first time. Firstly, the features of the protein sequence are extracted by using Chou’s pseudo amino acid composition (PseAAC). Then the extracted feature information is denoised by two-dimensional (2D) wavelet. Finally, the optimal feature vectors are input to support vector machine (SVM) classifier to predict protein structural classes. We obtained significant predictive results using jackknife test on three low-similarity protein structural class datasets 25PDB, 1189 and 640, and compared our method with previous methods The results indicate that the method proposed in this paper can effectively improve the prediction accuracy of protein structural class, which will be a reliable tool for prediction of protein structural class, especially for low-similarity sequences.     

A GMDH polynomial neural network-based method to predict approximate three-dimensional structures of polypeptides

Tertiary Protein Structure Prediction is one of the most important problems in Structural Bioinformatics. Along the last 20 years many algorithms have been proposed as to solve this problem. However, it still remains a challenging issue because of the complexity and of the dimensionality of the protein conformational search space. In this article a first principle method which uses database information for the prediction of the 3-D structure of polypeptides is presented. The technique is based on the Group Method of Data Handling (GMDH) algorithm, implemented by a software tool introduced on this work. GMDH Polynomial Neural Networks have been used with success in many fields such as data mining, knowledge discovery, pattern recognition and prediction. The proposed method was tested with seven protein sequences whose sizes vary from 14 to 54 amino acid residues. Results show that the predicted tertiary structures adopt a fold similar to the experimental structures. RMSD and secondary structure analysis reveal that the proposed method present accurate results in their predictions. The predicted structures can be used as input structures in refinement methods based on molecular mechanics (MM), e.g. molecular dynamics (MD) simulations. The search space is expected to be greatly reduced and the ab initio methods can demand a much reduced computational time to achieve a more accurate polypeptide structure.     

一种新的RNA二级结构特征序列表示及相似性分析*

针对目前不同的RNA二级结构可能对应相同特征序列的问题,本文提出了一种新的RNA二级结构特征序列表示法,根据不同的RNA二级结构的子结构类型,分别给出相应的字符表示并由此得到新的特征序列。利用Lempel-Ziv复杂度对两组RNA二级结构进行了相似性分析,实验结果表明了该表示法可以有效的提取RNA二级结构的结构信息,避免了不同RNA二级结构可能对应相同特征序列的问题。     

Toy模型蛋白质折叠问题的随机扰动粒子群解法

Toy模型蛋白质折叠问题是一个计算生物学中典型的NP难题。提出了一种随机扰动粒子群结合爬山优化的算法,应用二维Toy模型进行蛋白质折叠结构预测,在Fibonacci测试序列及真实蛋白质序列上的测试结果验证了算法的良好性能。     

基于散列辞典的蛋白质二级结构预测方法

提出一种易于修改的蛋白质二级结构预测算法。以蛋白质数据银行中PDB文本数据作为数据源,提取所有蛋白质氨基酸序列并以此建立样本数据库,然后针对α-螺旋、β-折叠分别利用基于散列辞典的不同改进方法编程实现蛋白质二级结构序列片段预测,在预测过程中,随机抽取68421个蛋白质中部分样本作为测试集,对未知序列根据建立的散列辞典中的片段使用正向最大匹配分词法进行切分对比。从实验结果来看,对未知序列片段预测的准确度达到了83．9％,而且能够较好地体现片段之间的连接顺序。     

利用自适应时间片梯度谱对镜头转换进行精细识别

提出了一种利用视频序列的自适应时间片梯度谱对镜头转换类型进行精细识别的算法 .在这个算法中 ,首先提出了自适应时间片梯度谱的概念 ,并且在此基础上 ,提出了用于识别镜头转换类型的模板构造和匹配算法 ,同时还给出了一种描述语言作为模板的软描述方法 ,用以适应不同的图象尺寸和镜头转换的持续时间 .实验结果表明 ,这种方法在多数情况下具有很好的识别效果 ,只是在两个镜头转换非常相近时 ,识别效果差些 ,另外 ,该方法识别速度快 ,同时具有很强的鲁棒性和可扩展性 ,是解决镜头转换类型精细识别问题的一个有益尝试     

基于结构比对的蛋白结合位点预测方法

蛋白质通过结合位点与其他分子产生相互作用, 所以对蛋白结合位点的预测具有重要的意义. 现有许多不同的预测方法, 但是这些方法存在命中率低或计算量大的问题, 本文引入了一种基于结构比对的蛋白质位点预测方法, 同时在结构比对过程中引入同源索引, 找出相应的同源模版, 并与之进行结构比对, 然后将结构相似的模版中的配体映射到目标蛋白质中, 采用聚类方法对位点进行分析. 结果表明, 与其他预测方法相比, 本文的方法降低了计算量, 并提高了预测精度.     

一种基于 Transformer 的三维人体姿态估计方法

三维人体姿态估计是人类行为理解的基础,但是预测出合理的三维人体姿态序列仍然是具有挑 战性的问题。为了解决这个问题,提出一种基于 Transformer 的三维人体姿态估计方法,利用多层长短期记忆 (LSTM)单元和多尺度 Transformer 结构增强人体姿态序列预测的准确性。首先,设计基于时间序列的生成器, 通过 ResNet 预训练神经网络提取图像特征;其次,采用多层 LSTM 单元学习时间连续性的图像序列中人体姿 态之间的关系,输出合理的 SMPL 人体参数模型序列;最后,构建基于多尺度 Transformer 的判别器,利用多 尺度 Transformer 结构对多个分割粒度进行细节特征学习,尤其是 Transformer block 对相对位置进行编码增强 局部特征学习能力。实验结果表明,该方法相对于 VIBE 方法具有更好地预测精度,在 3DPW 数据集上比 VIBE 的平均(每)关节位置误差(MPJPE)低了 7.5%;在 MP-INF-3DHP 数据集上比 VIBE 的 MPJPE 降低了 1.8%。      

一种序列相关支持向量机的B桶状跨膜蛋白预测方法

膜蛋白是一种具有重要生物功能的蛋白质,根据蛋白质的序列信息预测其是否属于β桶状跨膜蛋白是结构预测与功能分析的重要先导步骤,也是蛋白质预测领域中的一个挑战性问题。针对这两类问题,提取了208条β桶状跨膜蛋白序列的氨基酸位置与理化特征。利用支持向量机(SVM)进行了预测,结果表明二分类精度与相关系数分别达到了88.36%与0.7723。     

求解HP模型蛋白质折叠问题的启发式算法

构造了一个新的数学模型，把三维HP模型的蛋白质折叠问题由一个有约束的优化问题转化为无约束的优化问题，通过建立相对坐标和邻域结构，提出了一个局部搜索算法，并对文献中的链长不同的7个算例进行了测试。结果表明，该算法能在较短时间内找到其中5个算例的最优能量枸形，对另外2个难例，则可以找到能量仅比最优构形高一个单位的次优构形。     

An adaptive neural network strategy for improving the computational performance of evolutionary structural optimization

This work is focused on improving the computational efficiency of evolutionary algorithms implemented in large-scale structural optimization problems. Locating optimal structural designs using evolutionary algorithms is a task associated with high computational cost, since a complete finite element (FE) analysis needs to be carried out for each parent and offspring design vector of the populations considered. Each of these FE solutions facilitates decision making regarding the feasibility or infeasibility of the corresponding structural design by evaluating the displacement and stress constraints specified for the structural problem at hand. This paper presents a neural network (NN) strategy to reliably predict, in the framework of an evolution strategies (ES) procedure for structural optimization, the feasibility or infeasibility of structural designs avoiding computationally expensive FE analyses. The proposed NN implementation is adaptive in the sense that the utilized NN configuration is appropriately updated as the ES process evolves by performing NN retrainings using information gradually accumulated during the ES execution. The prediction capabilities and the computational advantages offered by this adaptive NN scheme coupled with domain decomposition solution techniques are investigated in the context of design optimization of skeletal structures on both sequential and parallel computing environments.     

Comparative protein modeling of methionine S-adenosyltransferase (MAT) enzyme from Mycobacterium tuberculosis: a potential target for antituberculosis drug discovery

Mycobacterium tuberculosis (Mtb) is a successful pathogen that overcomes the numerous challenges presented by the immune system of the host. In the last 40 years few anti-TB drugs have been developed, while the drug-resistance problem is increasing; there is thus a pressing need to develop new anti-TB drugs active against both the acute and chronic growth phases of the mycobacterium. Methionine S-adenosyltransferase (MAT) is an enzyme involved in the synthesis of S-adenosylmethionine (SAM), a methyl donor essential for mycolipid biosynthesis. As an anti-TB drug target, Mtb-MAT has been well validated. A homology model of MAT has been constructed using the X-ray structures of E. coli MAT (PDB code: 1MXA) and rat MAT (PDB code: 1QM4) as templates, by comparative protein modeling principles. The resulting model has the correct stereochemistry as gauged from the Ramachandran plot and good three-dimensional (3D) structure compatibility as assessed by the Profiles-3D score. The structurally and functionally important residues (active site) of Mtb-MAT have been identified using the E. coli and rat MAT crystal structures and the reported point mutation data. The homology model conserves the topological and active site features of the MAT family of proteins. The differences in the molecular electrostatic potentials (MEP) of Mtb and human MAT provide evidences that selective and specific Mtb-MAT inhibitors can be designed using the homology model, by the structure-based drug design approaches.     

Discriminating protein structure classes by incorporating Pseudo Average Chemical Shift to Chou's general PseAAC and Support Vector Machine

Proteins control all biological functions in living species. Protein structure is comprised of four major classes including all-α class, all-β class, α+β, and α/β. Each class performs different function according to their nature. Owing to the large exploration of protein sequences in the databanks, the identification of protein structure classes is difficult through conventional methods with respect to cost and time. Looking at the importance of protein structure classes, it is thus highly desirable to develop a computational model for discriminating protein structure classes with high accuracy. For this purpose, we propose a silco method by incorporating Pseudo Average Chemical Shift and Support Vector Machine. Two feature extraction schemes namely Pseudo Amino Acid Composition and Pseudo Average Chemical Shift are used to explore valuable information from protein sequences. The performance of the proposed model is assessed using four benchmark datasets 25PDB, 1189, 640 and 399 employing jackknife test. The success rates of the proposed model are 84.2%, 85.0%, 86.4%, and 89.2%, respectively on the four datasets. The empirical results reveal that the performance of our proposed model compared to existing models is promising in the literature so far and might be useful for future research.     

High throughput processing of the structural information in the protein data bank

The protein data bank (PDB) is the largest, most comprehensive, freely available depository of protein structural information, containing more than 37 500 deposited structures. On one hand, the form and the organization of the PDB seems to be perfectly adequate for gathering information from specific protein structures, by using the bibliographic references and the informative remark fields. On the other hand, however, it seems to be impossible to automatically review remark fields and journal references for processing hundreds or thousands of PDB files.

We present here a family of combinatorial algorithms to solve some of these problems. Our algorithms are capable to automatically analyze PDB structural information, identify missing atoms, repair chain ID information, and most importantly, the algorithms are capable of identifying ligands with their respective binding sites.     

An Algorithm for Simultaneous Backbone Threading and Side-Chain Packing

To utilize fully all available information in protein structure prediction, including both backbone and side-chain structures, we present a novel algorithm for solving a generalized threading problem. In this problem we consider simultaneous backbone threading and side-chain packing during the process of a protein structure prediction. For a given query protein sequence and a template structure, our goal is to find a threading alignment between the query sequence and the template structure, along with a rotamer assignment for each side-chain of the query protein, which optimizes an energy function that combines a backbone threading energy and a side-chain packing energy. This highly computationally challenging problem is solved through first formulating this problem as a graph-based optimization problem. Various graph-theoretic techniques are employed to achieve the computational efficiency to make our algorithm practically useful, which takes advantage of a number of special properties of the graph representing this generalized threading problem. The overall framework of our algorithm is a dynamic programming algorithm implemented on an optimal tree decomposition of the graph representation of our problem. By using various additional heuristic techniques such as dead-end elimination, we have demonstrated that our algorithm can solve a generalized threading problem within a practically acceptable amount of time and space, the first of its kind.     

An Algorithm for Simultaneous Backbone Threading and Side-Chain Packing

To fully utilize all available information in protein structure prediction, including both backbone and side-chain structures, we present a novel algorithm for solving a generalized threading problem. In this problem, we consider simultaneously backbone threading and side-chain packing during the process of a protein structure prediction. For a given query protein sequence and a template structure, our goal is to find a threading alignment between the query sequence and the template structure, along with a rotamer assignment for each side-chain of the query protein, which optimizes an energy function that combines a backbone threading energy and a side-chain packing energy. This highly computationally challenging problem is solved through first formulating this problem as a graph-based optimization problem. Various graph-theoretic techniques are employed to achieve the computational efficiency to make our algorithm practically useful, which takes advantage of a number of special properties of the graph representing this generalized threading problem. The overall framework of our algorithm is a dynamic programming algorithm implemented on an optimal tree decomposition of the graph representation of our problem. By using various additional heuristic techniques such as the dead-end elimination, we have demonstrated that our algorithm can solve a generalized threading problem within practically acceptable amount of time and space, the first of its kind.     

CLiBE: a database of computed ligand binding energy for ligand-receptor complexes

Consideration of binding competitiveness of a drug candidate against natural ligands and other drugs that bind to the same receptor site may facilitate the rational development of a candidate into a potent drug. A strategy that can be applied to computer-aided drug design is to evaluate ligand-receptor interaction energy or other scoring functions of a designed drug with that of the relevant ligands known to bind to the same binding site. As a tool to facilitate such a strategy, a database of ligand-receptor interaction energy is developed from known ligand-receptor 3D structural entries in the Protein Databank (PDB). The Energy is computed based on a molecular mechanics force field that has been used in the prediction of therapeutic and toxicity targets of drugs. This database also contains information about ligand function and other properties and it can be accessed at http://xin.cz3.nus.edu.sg/group/CLiBE.asp. The computed energy components may facilitate the probing of the mode of action and other profiles of binding. A number of computed energies of some PDB ligand-receptor complexes in this database are studied and compared to experimental binding affinity. A certain degree of correlation between the computed energy and experimental binding affinity is found, which suggests that the computed energy may be useful in facilitating a qualitative analysis of drug binding competitiveness.     

Robust topology optimization of structures with uncertainties in stiffness – Application to truss structures

A computational strategy is proposed for robust structural topology optimization in the presence of uncertainties with known second order statistics. The strategy combines deterministic topology optimization techniques with a perturbation method for the quantification of uncertainties associated with structural stiffness, such as uncertain material properties and/or structure geometry. The use of perturbation transforms the problem of topology optimization under uncertainty to an augmented deterministic topology optimization problem. This in turn leads to significant computational savings when compared with Monte Carlo-based optimization algorithms which involve multiple formations and inversions of the global stiffness matrix. Examples from truss structures are presented to show the importance of including the effect of controlling the variability in the final design. It is also shown that results obtained from the proposed method are in excellent agreement with those obtained from a Monte Carlo-based optimization algorithm.     

Improving Structural Statistical Machine Translation for Sign Language With Small Corpus Using Thematic Role Templates as Translation Memory

This paper presents a structural statistical machine translation (SSMT) model to deal with the data sparseness problem that occurs as a result of the necessarily small corpus to translate Chinese into Taiwanese Sign Language (TSL). A parallel bilingual corpus was developed, and linguistic information from the Sinica Treebank is adopted for Chinese sentence analysis. The synchronous context free grammar (SCFG) was adopted to convert a Chinese structure to the corresponding TSL structure and then extract a translation memory which comprises the thematic relations between the grammar rules of both structures. In structural translation, the statistical MT (SMT) approach was used to align the thematic roles in the grammar rules and the translation memory provides the reference templates for TSL structure translation. Finally, the agreement information for TSL verbs was labeled for enriching the expressiveness of the translated TSL sequence. Several experiments were conducted to evaluate the translation performance and the communication effectiveness for the deaf. The evaluation results demonstrate that the proposed approach outperforms a baseline statistical MT system using the same small corpus, especially for the translation of long sentences.