一个识别蛋白质折叠模式的SVM分类器

蛋白质折叠模式识别是一种分析蛋白质结构的重要方法。以序列相似性较低的蛋白质为训练集,提取蛋白质序列信息频数及疏水性等信息作为折叠类型特征,从SCOP数据库中已分类蛋白质构建1 393种折叠模式的数据集,采用SVM预测蛋白质1 393种折叠模式。封闭测试准确率达99.612 2%,基于SCOP的开放测试准确率达79.632 9%。基于另一个权威测试集的开放测试折叠准确率达64.705 9%,SCOP类准确率达76.470 6%,可以有效地对蛋白质折叠模式进行预测,从而为蛋白质从头预测提供参考。     

用人工神经网络方法预测蛋白质超二级结构

蛋白质超二级结构,即由α－螺旋和β－折叠等二级结构单元和连接短肽组成的超二级结构,是蛋白质结构研究中的一个重要层次。目前蛋白质超二级结构的预测工作尚属摸索阶段,还没有成熟的方法。人工神经网络预测方法是近年来在二级结构预测中发展起来的新方法。本文成功的将人工神经网络引入蛋白质超二级结构的预测工作中,结果表明蛋白质的超二级结构的发生与其局域的氨基酸的序列模式有重要联系,可以由蛋白质的一级结构序列预测该     

基于打分矩阵的多类蛋白质折叠子的预测

依据蛋白质折叠子中氨基酸保守性,以氨基酸、氨基酸的极性、氨基酸的电性以及氨基酸的亲—疏水性为参数,从蛋白质的氨基酸序列出发,采用"一对多"的分类策略,通过构建打分矩阵和选取氨基酸序列模式片断,利用5种相似性打分函数对27类折叠子进行识别,最好的预测精度达到83.46%。结果表明,打分矩阵是预测多类蛋白质折叠子有效的方法。     

模拟蛋白质折叠过程的新算法研究

蛋白质折叠过程模拟是当前蛋白质研究领域的一个难点问题。针对这一问题,提出了一个描述蛋白质折叠过程的算法-拟蛇算法,并且从分子振荡和分子动力学理论两个方面来证明该算法的核心函数是可行和正确的。经过实验总结出所有蛋白质空间结构都可以通过两种类型函数构造出来,提出了描述蛋白质折叠过程模型。与其它蛋白质折叠过程模拟算法的实验结果比较表明,拟蛇算法所构造的空间结构能量值最小、相似度最好。进而说明拟蛇算法和蛋白质折叠过程模型在描述蛋白质折叠过程方面具有明显优势。     

基于序列疏水值震荡的折叠速率预测

蛋白质折叠速率的正确预测对理解蛋白质的折叠机理非常重要。本文从伪氨基酸组成的方法出发,提出利用序列疏水值震荡的方法来提取蛋白质氨基酸的序列顺序信息,建立线性回归模型进行折叠速率预测。该方法不需要蛋白质的任何二级结构、三级结构信息或结构类信息,可直接从序列对蛋白质折叠速率进行预测。对含有62个蛋白质的数据集,经过Jack．knife交互检验验证,相关系数达到0．804,表示折叠速率预测值与实验值有很好的相关性,说明了氨基酸序列信息对蛋白质折叠速率影响重要。同其他方法相比,本文的方法具有计算简单,输入参数少等特点。     

基于氨基酸约化和统计特征的蛋白质亚细胞定位预测

蛋白质亚细胞定位预测对蛋白质的功能、相互作用及调控机制的研究具有重要意义。本文基于物化性质和结构性质对氨基酸的约化,描述序列局部和全局信息的"组成"、"转换"和"分布"特征,并利用氨基酸亲疏水性的数值统计特征,提出了一种新的蛋白质特征表示方法(NSBH)。分别使用三种分类器KNN、SVM及BP神经网络进行蛋白质亚细胞定位预测,比较了几种方法和特征融合方法的预测结果,显示融合特征表示及结合SVM分类器时能够达到更好的预测准确率。同时,还详细讨论了不同参数对实验结果的影响,具体的实验及比较结果显示了该方法的有效性。     

结合支持向量机和贝叶斯方法进行蛋白质二级结构预测

组建一个分两个阶段的分类器来进行蛋白质二级结构预测。第一阶段由支持向量机分类器组成,在第二阶段中使用第一阶段已预测的结果来进行贝叶斯判别。预测性能的改进表明了结合支持向量机和贝叶斯方法预测性能优越于单独使用支持向量机的预测性能。同时也证明残基在形成二级结构时是相互影响的。     

蛋白质折叠识别方法综述

蛋白质折叠识别算法是蛋白质三维结构预测的重要方法之一,该方法在生物科学的许多方面得到卓有成效的应用。在过去的十年中,我们见证了一系列基于不同计算方式的蛋白质折叠识别方法。在这些计算方法中,机器学习和序列谱-序列谱比对是两种在蛋白质折叠中应用较为广泛和有效的方法。除了计算方法的进展外,不断增大的蛋白质结构数据库也是蛋白质折叠识别的预测精度不断提高的一个重要因素。在这篇文章中,我们将简要地回顾蛋白质折叠中的先进算法。另外,我们也将讨论一些可能可以应用于改进蛋白质折叠算法的策略。     

基于模糊支持向量机的膜蛋白折叠类型预测

现有的基于支持向量机（support vector machine,SVM）来预测膜蛋白折叠类型的方法．利用的蛋白质序列特征并不充分．并且在处理多类蛋白质分类问题时存在不可分区域,针对这两类问题．提取蛋白质序列的氨基酸和二肽组成特征,并计算加权的多阶氨基酸残基指数相关系数特征,将3类特征融和作为分类器的输入特征矢量．并采用模糊SVM（fuzzy SVM,FSVM）算法解决对传统SVM不可分数据的分类．在无冗余的数据集上测试结果显示．改进的特征提取方法在相同分类算法下预测性能优于已有的特征提取方法：FSVM在相同特征提取方法下性能优于传统的SVM．二者相结合的分类策略在独立性数据集测试下的预测精度达到96．6％．优于现有的多种预测方法．能够作为预测膜蛋白和其它蛋白质折叠类型的有效工具．     

对预测蛋白质结构的拟物算法的分析和改进

对预测蛋白质空间结构的拟物算法的有效性进行理论分析,证明用该拟物算法求得合法的结构存在较大的随机性;给出折叠结构发生冲突的判断条件和提高拟物算法有效性的一些修正方案。     

Ensemble classifier for protein fold pattern recognition

MOTIVATION: Prediction of protein folding patterns is one level deeper than that of protein structural classes, and hence is much more complicated and difficult. To deal with such a challenging problem, the ensemble classifier was introduced. It was formed by a set of basic classifiers, with each trained in different parameter systems, such as predicted secondary structure, hydrophobicity, van der Waals volume, polarity, polarizability, as well as different dimensions of pseudo-amino acid composition, which were extracted from a training dataset. The operation engine for the constituent individual classifiers was OET-KNN (optimized evidence-theoretic k-nearest neighbors) rule. Their outcomes were combined through a weighted voting to give a final determination for classifying a query protein. The recognition was to find the true fold among the 27 possible patterns. RESULTS: The overall success rate thus obtained was 62% for a testing dataset where most of the proteins have <25% sequence identity with the proteins used in training the classifier. Such a rate is 6-21% higher than the corresponding rates obtained by various existing NN (neural networks) and SVM (support vector machines) approaches, implying that the ensemble classifier is very promising and might become a useful vehicle in protein science, as well as proteomics and bioinformatics. AVAILABILITY: The ensemble classifier, called PFP-Pred, is available as a web-server at http://202.120.37.186/bioinf/fold/PFP-Pred.htm for public usage.     

基于添加模体信息和功率谱密度的组合向量预测27类蛋白质折叠子

以序列相似性低于40%的1895条蛋白质序列构建涵盖27个折叠类型的蛋白质折叠子数据库,从蛋白质序列出发,用模体频数值、低频功率谱密度值、氨基酸组分、预测的二级结构信息和自相关函数值构成组合向量表示蛋白质序列信息,采用支持向量机算法,基于整体分类策略,对27类蛋白质折叠子的折叠类型进行预测,独立检验的预测精度达到了66.67%。同时,以同样的特征参数和算法对27类折叠子的4个结构类型进行了预测,独立检验的预测精度达到了89.24%。将同样的方法用于前人使用过的27类折叠子数据库,得到了好于前人的预测结果。     

Predicting protein fold pattern with functional domain and sequential evolution information

The fold pattern of a protein is one level deeper than its structural classification, and hence is more challenging and complicated for prediction. Many efforts have been made in this regard, but so far all the reported success rates are still under 70%, indicating that it is extremely difficult to enhance the success rate even by 1% or 2%. To address this problem, here a novel approach is proposed that is featured by combining the functional domain information and the sequential evolution information through a fusion ensemble classifier. The predictor thus developed is called PFP-FunDSeqE. Tests were performed for identifying proteins among their 27 fold patterns. Compared with the existing predictors tested by a same stringent benchmark dataset, the new predictor can, for the first time, achieve over 70% success rate. The PFP-FunDSeqE predictor is freely available to the public as a web server at http://www.csbio.sjtu.edu.cn/bioinf/PFP-FunDSeqE/.     

Influence of medium- and long-range interactions in different folding types of globular proteins

Recognition of protein fold from amino acid sequence is a challenging task. The structure and stability of proteins from different fold are mainly dictated by inter-residue interactions. In our earlier work, we have successfully used the medium- and long-range contacts for predicting the protein folding rates, discriminating globular and membrane proteins and for distinguishing protein structural classes. In this work, we analyze the role of inter-residue interactions in commonly occurring folds of globular proteins in order to understand their folding mechanisms. In the medium-range contacts, the globin fold and four-helical bundle proteins have more contacts than that of DNA-RNA fold although they all belong to all-alpha class. In long-range contacts, only the ribonuclease fold prefers 4-10 range and the other folding types prefer the range 21-30 in alpha/beta class proteins. Further, the preferred residues and residue pairs influenced by these different folds are discussed. The information about the preference of medium- and long-range contacts exhibited by the 20 amino acid residues can be effectively used to predict the folding type of each protein.     

Influence of denatured and intermediate states of folding on protein aggregation

We simulate the aggregation thermodynamics and kinetics of proteins L and G, each of which self-assembles to the same alpha/beta [corrected] topology through distinct folding mechanisms. We find that the aggregation kinetics of both proteins at an experimentally relevant concentration exhibit both fast and slow aggregation pathways, although a greater proportion of protein G aggregation events are slow relative to those of found for protein L. These kinetic differences are correlated with the amount and distribution of intrachain contacts formed in the denatured state ensemble (DSE), or an intermediate state ensemble (ISE) if it exists, as well as the folding timescales of the two proteins. Protein G aggregates more slowly than protein L due to its rapidly formed folding intermediate, which exhibits native intrachain contacts spread across the protein, suggesting that certain early folding intermediates may be selected for by evolution due to their protective role against unwanted aggregation. Protein L shows only localized native structure in the DSE with timescales of folding that are commensurate with the aggregation timescale, leaving it vulnerable to domain swapping or nonnative interactions with other chains that increase the aggregation rate. Folding experiments that characterize the structural signatures of the DSE, ISE, or the transition state ensemble (TSE) under nonaggregating conditions should be able to predict regions where interchain contacts will be made in the aggregate, and to predict slower aggregation rates for proteins with contacts that are dispersed across the fold. Since proteins L and G can both form amyloid fibrils, this work also provides mechanistic and structural insight into the formation of prefibrillar species.     

Decision tree based information integration for automated protein classification

We propose a novel technique for automatically generating the SCOP classification of a protein structure with high accuracy. We achieve accurate classification by combining the decisions of multiple methods using the consensus of a committee (or an ensemble) classifier. Our technique, based on decision trees, is rooted in machine learning which shows that by judicially employing component classifiers, an ensemble classifier can be constructed to outperform its components. We use two sequence- and three structure-comparison tools as component classifiers. Given a protein structure and using the joint hypothesis, we first determine if the protein belongs to an existing category (family, superfamily, fold) in the SCOP hierarchy. For the proteins that are predicted as members of the existing categories, we compute their family-, superfamily-, and fold-level classifications using the consensus classifier. We show that we can significantly improve the classification accuracy compared to the individual component classifiers. In particular, we achieve error rates that are 3-12 times less than the individual classifiers' error rates at the family level, 1.5-4.5 times less at the superfamily level, and 1.1-2.4 times less at the fold level.     

Pathway heterogeneity in protein folding

We generate ab initio folding pathways in two single-domain proteins, hyperthermophile variant of protein G domain (1gb4) and ubiquitin (1ubi), both presumed to be two-state folders. Both proteins are endowed with the same topology but, as shown in this work, rely to a different extent on large-scale context to find their native folds. First, we demonstrate a generic feature of two-state folders: A downsizing of structural fluctuations is achieved only when the protein reaches a stationary plateau maximizing the number of highly protected hydrogen bonds. This enables us to identify the folding nucleus and show that folding does not become expeditious until a topology is generated that is able to protect intramolecular hydrogen bonds from water attack. Pathway heterogeneity is shown to be dependent on the extent to which the protein relies on large-scale context to fold, rather than on contact order: Proteins that can only stabilize native secondary structure by packing it against scaffolding hydrophobic moieties are meant to have a heterogeneous transition-state ensemble if they are to become successful folders (otherwise, successful folding would be too fortuitous an event.) We estimate mutational Phi values as ensemble averages and deconvolute individual-route contributions to the averaged two-state kinetic picture. Our results find experimental corroboration in the well-studied chymotrypsin inhibitor (CI2), while leading to verifiable predictions for the other two study cases.     

Structural comparison of the two alternative transition states for folding of TI I27

TI I27, a beta-sandwich domain from the human muscle protein titin, has been shown to fold via two alternative pathways, which correspond to a change in the folding mechanism. Under physiological conditions, TI I27 folds by a classical nucleation-condensation mechanism (diffuse transition state), whereas at extreme conditions of temperature and denaturant it switches to having a polarized transition state. We have used experimental Phi-values as restraints in ensemble-averaged molecular dynamics simulations to determine the ensembles of structures representing the two transition states. The comparison of these ensembles indicates that when native interactions are substantially weakened, a protein may still be able to fold if it can access an alternative transition state characterized by a much larger entropic contribution. Analysis of the probability distribution of Phi-values derived from ensemble averaged simulations, enables us to identify residues that form contacts in some members of the ensemble but not in others illustrating that many interactions present in transition states are not strictly required for the successful completion of the folding process.     

Transition state contact orders correlate with protein folding rates

We have used molecular dynamics simulations restrained by experimental phi values derived from protein engineering experiments to determine the structures of the transition state ensembles of ten proteins that fold with two-state kinetics. For each of these proteins we then calculated the average contact order in the transition state ensemble and compared it with the corresponding experimental folding rate. The resulting correlation coefficient is similar to that computed for the contact orders of the native structures, supporting the use of native state contact orders for predicting folding rates. The native contacts in the transition state also correlate with those of the native state but are found to be about 30% lower. These results show that, despite the high levels of heterogeneity in the transition state ensemble, the large majority of contributing structures have native-like topologies and that the native state contact order captures this phenomenon.     

Thermodynamics and folding kinetics analysis of the SH3 domain form discrete molecular dynamics

We perform a detailed analysis of the thermodynamics and folding kinetics of the SH3 domain fold with discrete molecular dynamic simulations. We propose a protein model that reproduces some of the experimentally observed thermodynamic and folding kinetic properties of proteins. Specifically, we use our model to study the transition state ensemble of the SH3 fold family of proteins, a set of unstable conformations that fold to the protein native state with probability 1/2. We analyze the participation of each secondary structure element formed at the transition state ensemble. We also identify the folding nucleus of the SH3 fold and test extensively its importance for folding kinetics. We predict that a set of amino acid contacts between the RT-loop and the distal hairpin are the critical folding nucleus of the SH3 fold and propose a hypothesis that explains this result.