基于二叉树支持向量机的蛋白质结构类预测

提出了一种基于二叉树支持向量机(BT-SVM)的蛋白质结构类多类预测新方法.采用26维的向量来表示蛋白质序列的特征.BT-SVM多类分类方法能消除SVM在多分类问题中存在的不可分数据问题.采用两个经典数据集作为测试数据,通过自身一致性和n折叠交叉验证方法测试了新方法的性能.预测结果表明新方法具有良好的预测能力,与使用同一数据集的已有结果相比较,新方法的Jackknife结果和目前最好的方法取得的结果相当,可作为蛋白质结构类预测的一个工具.     

基于多特征信息及Ma-Ada多分类器融合的蛋白质结构类预测

蛋白质序列特征表示和机器学习算法是影响蛋白质结构类预测效果好坏的两个重要方面.本研究基于k-字统计频率和k-片段位置分布两种特征提取方法,将分别提取到的氨基酸序列信息和物理化学性质信息同蛋白质二级结构信息进行融合,建立17维和57维的特征信息集,并尝试在Adaboost.M1算法中引入Multi-Agent多智能体融合的思想,提出了一种Ma-Ada多分类器融合算法.该算法作为蛋白质结构类的预测工具,充分挖掘了单分类器度量层信息以及各个单分类器之间的交互融合信息.实验结果表明,Ma-Ada算法在Z277、Z498、1189和D640四个蛋白质数据集的57维特征信息集上的分类率分别达到了91.3％、96.8％、85.3％和87.2％,在17维特征信息集上的分类率也分别达到了90.6％、95.8％、84.8％和88.3％.与其它蛋白质结构类预测方法的结果相比,本方法能够获得较好的分类率.     

基于最大熵模型的蛋白质二级结构的预测

蛋白质二级结构预测是蛋白质功能预测中一个很重要的环节.我们综合考虑了蛋白质序列中对氨基酸二级结构具有影响的因素,采用最大熵模型实现了对蛋白质二级结构的预测.将氨基酸之间的相互作用和氨基酸对于二级结构的倾向性作为特征函数,充分体现了蛋白质序列中氨基酸之间的短程和长程的相互作用的影响.最大熵模型整合了这些多源的信息,形成一个单一的概率模型.与现有的单序列预测方法相比,本文的方法取得了较高的预测准确率,预测结果表明它是一种实用的预测方法.     

结合蛋白质互作与功能类的可分性预测蛋白质功能

当前蛋白质功能注释体系的欠完备性限制了生物学及医药研究的进一步发展和应用,有必要将基因本体功能知识体系 (GO)的功能注释信息进一步深化,把蛋白质注释到GO中更具体的功能节点.为此,提出一种结合互作信息的新的预测策略,将酵母蛋白质准确地预测到更具体的功能类中.针对GO中的每个候选预测空间来构建分类器,并选用功能类分离性指标对候选预测空间进行评价,选出该指标大于一定阈值的候选预测空间,再将父节点中的蛋白质预测到子节点中.通过扩展深化预测的范围,可将预测空间一直上溯到根节点,对蛋白质功能进行深层预测,得到很好的预测结果.以上溯两层的预测空间为例,平均真阳性率和覆盖率分别达到94.02%和95.82%.     

基于氨基酸疏水性与PSSM构建ANN预测蛋白质二级结构

目的 预测蛋白质二级结构是预测其空间结构的基础,提高蛋白质二级结构的预测率非常重要.方法在本研究中,结合氨基酸的疏水性与含有进化信息的位置特异性得分矩阵(PSSM),构建BP神经网络.本文的数据来源于蛋白质数据集合CB513,在此集合中去除氨基酸个数小于30及含有X、B的序列,共492条蛋白序列作为数据集.通过4-交互验证预测准确率.在本研究中,将蛋白质二级结构预测的结果与仅用PSSM作为输入的神经网络预测相比较.结果 采用疏水性与进化信息相结合作为输入所构建的神经网络对α螺旋的预测准确率有了较大的提高,达到近79%,敏感性及特异性分别达到79%及91%.同时对二级结构总体预测准确率达到75.96%.结论 此种方法构建的BP网络能提高蛋白质二级结构,尤其是α螺旋的预测准确率.     

基于次生特征提取方法预测蛋白质同源寡聚体

寡聚蛋白质相对于单体蛋白质具有许多优势,广泛地参与多种生命活动。本文提出次生特征提取方法,使用支持向量机作为分类器,采用"一对一"的多类分类策略,基于蛋白质一级序列提取特征方法,对四类同源寡聚体进行分类研究。结果表明,在Jackknife检验下,基于次生特征和氨基酸组成成分特征构成的特征集,加权情况下,其总分类精度最高达到了78.41%,比氨基酸组成成分特征提高13.09%,比参考文献最好特征集BG提高了6.86%,比最好原生特征集CM1提高了5.53%。此结果说明次生特征提取方法对于蛋白质同源寡聚体分类是一种非常有效的特征提取方法。     

基于局部支持向量机的蛋白质相互作用的预测方法

针对蛋白质相互作用的预测问题,我们提出了一种基于局部支持向量机的预测方法。该方法充分考虑了蛋白质相互作用数据的局部相似性特征,提出在待测样本附近构建支持向量机模型。对两个真实的蛋白质相互作用数据集H.pylori和Human的测试表明,基于局部支持向量机的预测方法能够有效剔除无用样本对待测样本的负面影响有效地提高了蛋白质相互作用预测的性能与其它方法相比具有一定的优势。     

基于MATLAB的隐马尔可夫模型预测蛋白质结构类

目的 准确预测蛋白质结构类,为研究其空间结构及生物功能打下基础.方法 应用隐马尔可夫模型(HMM)预测蛋白质结构类,分别构建3-状态HMM和8-状态HMM.数据来源于Chou和Zhou构建的蛋白质数据集,分别包含有204条蛋白质序列和498条蛋白质序列,通过留一法预测其准确率.结果 所构建的3-状态HMM和8-状态HMM对全α类的预测准确率最高,尤其是3-状态HMM的预测准确率达到95％以上.与Chou数据集相比,Zhou数据集对于全β类和α/β类的预测准确率也有所提高,同时,总体预测率也提高了2％左右;但α+β类的预测准确率有所下降.结论 将整条蛋白质序列作为预测模型的输入信息所构建的HMM模型能有效地预测蛋白质的结构类.     

从肿瘤基因表达数据挖掘分类规则的研究

基于肿瘤基因表达数据,利用信息科学的方法和技术建立肿瘤预测分类模型,对肿瘤基因表达模式研究和肿瘤的诊断识别具有重要意义.本研究提出一种从肿瘤基因表达数据中直接挖掘分类规则建立肿瘤预测分类器的方法.该方法首先抽取实验样本集,分别找出标记肿瘤和正常组织样本的分类特征,由此生成可预测样本类别的分类规则,对每个未知类别样本,按照置信度最高原则,选择一个分类规则作为预测结构.本研究的实验数据来自Broad Institute的前列腺癌基因表达数据,实验结果显示该方法的预测精度在90%以上,且同时获得了大量结构透明的分类预测规则,表明本研究的方法是可行的和有效的.     

基于模糊支持向量机的膜蛋白分类研究

以氨基酸组成为特征对膜蛋白的分类,忽略了序列残基之间的相关性信息,而采用传统支持向量机算法作为分类算法,在解决多类问题时会出现分类盲区问题。针对这两种情况,计算蛋白质序列的氨基酸组成、二肽组成以及6种氨基酸相关系数,将三类特征结合,作为膜蛋白序列的特征向量;同时采用模糊支持向量机作为分类器,解决了传统支持向量机在多类数据识别中的盲区问题。测试结果表明,在相同特征输入下,模糊支持向量机分类性能优于传统支持向量机;在相同分类器的情况下,氨基酸组成、二肽组成和相关系数组合的特征选择方法的分类性能优于只使用其中一类或两类特征的方法;而采取组合特征和模糊支持向量机相结合的分类策略,在独立性数据集测试中的整体预测精度达到97%,优于现有的多种分类策略,是目前最有效的膜蛋白分类方法之一。     

Lymphokine-mediated fusion and migration inhibition of alveolar macrophages

Guinea pigs were shown to produce a lymphokine termed macrophage fusion factor (MFF) which mediated the fusion of 70–80% of guinea pig or rabbit alveolar macrophages, but not guinea pig peritoneal macrophages. In the conventional migration inhibitory factor (MIF) assay, guinea pig aveolar macrophages were inhibited in their migration and large numbers of giant cells were present. There appeared to be a correlation between the titer of MFF and migration inhibition of alveolar macrophages but not with MIF titer as expressed on the peritoneal macrophage. Guinea pig MFF production was erratic and its absence from lymphokine supernatant fluids correlated with an absence of migration inhibitory activity for the alveolar macrophage. Guinea pig MIF production was more constant and high titers were invariably present. Rabbit crude lymphokine supernatant fluids containing MFF also inhibited the migration of their alveolar macrophages when measured at 24 and 48 hr during the MIF assay. Extensive numbers of giant cells were observed in the cell fan whenever migration inhibition was present. α-l-Fucose, which is known to block the receptor sites of MIF, failed to block giant cell formation in either the MFF or the MIF assay and also failed to block migration inhibition of the alveolar macrophages. The results suggest that lymphokines other than MIF can inhibit the migration of alveolar macrophages in the standard MIF assay and that the lymphokine responsible for migration inhibition and fusion of alveolar macrophages is the same lymphokine, MFF.     

Glucocorticoid modulation of lymphokine-induced giant cell formation

Lymph node lymphocytes from rabbits sensitized with bacillus Calmet Guerin (BCG) secreted into the culture media both macrophage fusion factor (MFF) and migration inhibition factor (MIF) after 24 h of incubation with heat-killed BCG. Cell-free supernatant fluids obtained from these cultures induced simultaneously giant cell formation and migration inhibition of homologous normal alveolar macrophages. The glucocorticoids cortisol (10^–7 M) and dexa-methasone (10^–8 M) (DX) consistently inhibited giant cell formation elicited by MFF (P=0.003) without affecting macrophage viability. By contrast, the same glucocorticoids, in concentrations ranging from 10^–8 to 10^–10 M, induced a considerable increment of giant cell development in macrophage populations exhibiting a low response to MFF. Neither cortisol (10^–4 M) nor DX (10^–4 M) affected the migration inhibition of alveolar macrophages induced by MIF. Present results suggest that the granulomatous response in the rabbit, as reflected by the macrophage fusion assay, may be regulated by glucocorticoids.This paper was partially presented at the 8th International Convocation in Immunology. Buffalo, New York, June 14–17, 1982.     

Chemical Nature of the Interaction between Macrophage Fusion Factor and Macrophage Membranes

Giant-cell formation induced by macrophage fusion factor (MFF) was not altered after pretreatment of macrophages with trypsin, chymotrypsin, pronase, neuraminidase, phospholipase C, or phospholipase D. Pretreatment of macrophages with either alpha-mannosidase or alpha-glucosidase completely inhibited giant-cell development, without altering macrophage viability. No alteration of giant-cell formation was observed when 0.1 M of L-fucose, N-acetyl-glucosamine, D-arabinose, D-xylose, melibiose, D-glucose, D-galactose, alpha-lactose, sucrose, D-fructose, or maltose was present during incubation of macrophages with MFF. Giant-cell formation was abolished when 0.1 M alpha-D-mannose was present during macrophage incubation with MFF. These results suggest that the protein moiety of MFF recognizes a specific receptor site on the macrophage membrane, one that is different from those described for other lymphokines and contains alpha-mannose.     

Macrophage fusion factor elicited from BGG-sensitized lymphocytes.

Lymphocytes obtained from rabbit lymph nodes sensitized to bovine gamma globulin produce in vitro the lymphokine macrophage fusion factor (MFF) which mediates the fusion of approximately 100% of normal alveolar and oil-induced peritoneal macrophages. Giant cells (GC) of Langhans and foreign body type form large syncytia containing as many as several hundred nuclei per cell. Nuclei of GC appear more spherical and larger than those of the normal mononucleated macrophages, and they possess several prominent nucleoli. Giant cells of peritoneal macrophage origin show enhanced intracytoplasmic vacuolization. Normal macrophages cultured as a monolayer in MFF-rich supernatants form cell clusters which progressively fuse during the 24-hour incubation period. A signoid dose-response curve was obtained for cell fusion with MFF-rich supernatants possessing high titers, ie, the latter supernatants undiluted partially inhibited macrophage fusion. MIF-like activity was detected in MFF-rich supernatants as well as a factor(s) which inhibited 3H-thymidine uptake by giant cells.     

The effect of diethylcarbamazine on microfilariae ofLitomosoides carinii in vitro and in vivo

Culture-derivedLitomosoides carinii microfilariae (MFF) were used in in vitro and in vivo systems to investigate the effect of diethylcar-bamazine (DEC) on these MFF. In vivo: Male rats,Mastomys natalensis, all of the same age, were injected intrathoracically (12) or intraperitoneally (36) with 10³ or 10⁴ MFF. After 30 min one half of each group of rats was given DEC per os. At 30, 60, and 120 min after DEC administration, two rats from the treated and two from the untreated group were bled and killed. The pleural or peritoneal cavities were rinsed with warm saline (0.15m NaCl) to recover MFF. In both the intrathoracic and intraperitoneal experiments, equal numbers of MFF were recovered from treated and control rats at 30 and 120 min. However, at 60 min 85.5% fewer were recovered from the treated than from the nontreated animals. MFF were not found in the blood. In vitro: MFF were added to tissue culture dish wells (Linbro Div., Flow Labs, Hamden, Conn) prepared as follows: DEC-Serum (serum from normal rats given DEC at 500 mg/kg), DEC+Serum (serum with added DEC), serum only, RPMI 1640 only, and RPMI 1640+DEC. Furthermore, the five treatments were prepared either with or without unstimulated peritoneal exudate (PE) cells. At 30 min in the DEC-Serum wells 45% of the MFF had adherent PE cells; in the remaining wells these cells adhered to 11% or fewer MFF. We interpret the aforementioned phenomena as representing the first step in the trapping and elimination of MFF after DEC treatment ofL. carinii-infectedM. natalensis.     

Secretion of Macrophage Fusion Factor (MFF) by Schistosome Egg Granulomas Maintained In Vitro

The schistosome granuloma cultured in vitro under antigenic specific and unspecific stimulation releases macrophage fusion factor (MFF) into the medium, adjusting the production dynamics to the spontaneous modulation model already described. These findings demonstrate the feasibility of studying granulomatous response in mammalian schistosomiasis utilizing an in vitro model of lymphokine granuloma secretion.     

Characteristics of Epstein–Barr virus envelope protein gp42

Epstein–Barr virus (EBV) glycoprotein 42 (gp42) is a membrane protein essential for fusion and entry of EBV into host B-lymphocytes. Gp42 is a member of the protein-fold family C-type lectin or lectin-like domains (CLECT or CTLD) and specifically is classified as a natural-killer receptor (NKR)-like CLECT. Literature review and phylogenetic comparison show that EBV gp42 shares a common structure with other NKR-like CLECTs and possibly with many viral CTLDs, but does not appear to exhibit some common binding characteristics of many CTLDs, such as features required for calcium binding. The flexible N-terminal region adjacent to the CTLD fold is important for binding to other EBV glycoproteins and for a cleavage site that is necessary for infection of host cells. From structural studies of gp42 unbound and bound to receptor and extensive mutational analysis, a general model of how gp42 triggers membrane fusion utilizing both the flexible N-terminal region and the CTLD domain has emerged.     

FRAN and RBF-PSO as two components of a hyper framework to recognize protein folds

In this paper, an intelligent hyper framework is proposed to recognize protein folds from its amino acid sequence which is a fundamental problem in bioinformatics. This framework includes some statistical and intelligent algorithms for proteins classification. The main components of the proposed framework are the Fuzzy Resource-Allocating Network (FRAN) and the Radial Bases Function based on Particle Swarm Optimization (RBF-PSO). FRAN applies a dynamic method to tune up the RBF network parameters. Due to the patterns complexity captured in protein dataset, FRAN classifies the proteins under fuzzy conditions. Also, RBF-PSO applies PSO to tune up the RBF classifier. Experimental results demonstrate that FRAN improves prediction accuracy up to 51% and achieves acceptable multi-class results for protein fold prediction. Although RBF-PSO provides reasonable results for protein fold recognition up to 48%, it is weaker than FRAN in some cases. However the proposed hyper framework provides an opportunity to use a great range of intelligent methods and can learn from previous experiences. Thus it can avoid the weakness of some intelligent methods in terms of memory, computational time and static structure. Furthermore, the performance of this system can be enhanced throughout the system life-cycle.     

一个RNA剪接调控元件分类方法的研究

序列分类方法被广泛应用于各种生物信息学问题,例如转录调控元件识别和蛋白结构预测。本研究设计了一个新的基于序列特征的分类方法,并将其用于RNA剪接调控元件的研究。该方法从已知剪接元件中抽取序列特征,构建一个打分算法,由此预测未知元件RNA剪接调控功能。作为应用实例,采用已知外显子剪接增强子和沉默子(ESE和ESS)八联体作为实验数据,对本方法和若干已知常用方法的预测结果进行比较,3类计算验证实验中的平均预测精度为93%,表现出良好预测精度,且其透明的预测结构可帮助进行生物解释。该研究提供了一种可用于分析生物序列数据的新方法,给出了一个从生物信息学角度来研究基因调控问题的新途径。     

The Proteome Folding Project: proteome-scale prediction of structure and function

The incompleteness of proteome structure and function annotation is a critical problem for biologists and, in particular, severely limits interpretation of high-throughput and next-generation experiments. We have developed a proteome annotation pipeline based on structure prediction, where function and structure annotations are generated using an integration of sequence comparison, fold recognition, and grid-computing-enabled de novo structure prediction. We predict protein domain boundaries and three-dimensional (3D) structures for protein domains from 94 genomes (including human, Arabidopsis, rice, mouse, fly, yeast, Escherichia coli, and worm). De novo structure predictions were distributed on a grid of more than 1.5 million CPUs worldwide (World Community Grid). We generated significant numbers of new confident fold annotations (9% of domains that are otherwise unannotated in these genomes). We demonstrate that predicted structures can be combined with annotations from the Gene Ontology database to predict new and more specific molecular functions.