基于属性约简的PLS加权朴素贝叶斯分类

朴素贝叶斯算法是一种简单而高效的分类算法,它的属性独立性假设,影响了它的分类性能.针对这种问题,在分析属性相关性的基础上,通过属性约简选择一组近似独立的属性约简子集,提出一种基于属性约简的偏最小二乘回归加权朴素贝叶斯分类算法.对不同的条件属性给予不同的权值,从而在保持简单性的基础上有效地提高了朴素贝叶斯分类算法的分类性能.实验结果表明,该方法可行且有效.     

海洋底泥环境DNA提取与纯化方法比较研究

为了探索有效的海洋底泥宏基因组DNA提取与纯化方法,分别采用CTAB结合SDS法、CTAB结合玻璃珠法、SDS结合蛋白酶K法和SoilMasterTM试剂盒法提取海洋底泥宏基因组DNA,用柱式腐殖酸清除试剂盒法与电洗脱法纯化宏基因组DNA,并对宏基因组DNA的产量与纯度进行评价.结果表明:SDS结合蛋白酶K法提取率最高,DNA片段完整;SoilMasterTM试剂盒法与CTAB结合SDS法次之;CTAB结合玻璃珠法有严重降解;电洗脱法可以得到高纯度、大于42kb宏基因组DNA片段.因此,SDS结合蛋白酶K法和电洗脱法提取纯化海洋近海底泥宏基因组DNA优于其他方法.     

基于改进混合蛙跳算法的粗糙属性交叉熵优化约简

结合粗糙集属性约简二进制优化模型,提出一种基于改进混合蛙跳算法的粗糙属性交叉熵优化约简算法,该算法将粗糙集属性划分至不同蛙群进化模因组内,每个模因组内属性集设计成以精英个体为中心力的蛙群并行演化方式,并采用交叉熵最小原理进行精英个体寻优全局最优约简集,快速而有效地处理大规模信息系统的属性约简.UCI仿真实验结果表明本文提出的算法在搜索全局最小属性约简解效率和精度方面具有明显优势,该算法应用于含噪音的人脑核磁共振图像MRI分割实验,其对MRI图像分割的高效性进一步表明该算法具有较强的适用性.     

活性污泥宏基因组DNA提取方法优化

宏基因组DNA的成功获取是宏基因组学技术顺利开展的关键.利用溶菌酶-SDS-蛋白酶K法（LSK法）提取活性污泥宏基因组DNA,结合单因素法与表面响应法对提取过程的4个关键步骤进行条件优化,即预处理、细胞裂解、蛋白去除及DNA沉淀过程.确定影响DNA产量的重要因素：预处理缓冲溶液与DNA沉淀剂类型、CTAB质量分数、溶菌酶浓度、37℃水浴时间及SDS质量分数.确定LSK法提取宏基因组DNA的最佳条件：TENC预处理活性污泥;1.5%CTAB;0.5mg/mL溶菌酶,37℃水浴1.4h;2%SDS,200μg/mL蛋白酶,55℃水浴2.5h;异丙醇沉淀DNA 40min.在最佳条件下,活性污泥宏基因组DNA的最大产量为每g污泥170μg.本研究可为环境样品中高质量宏基因组DNA的提取提供有价值的参考.     

自来水中细菌宏基因组DNA的浓度与细菌菌落总数的相关性分析

分析自来水中细菌宏基因组DNA浓度与细菌菌落总数之间的相关性,探索新的自来水细菌菌落总数检测方法。采用国标法检测自来水中的细菌菌落总数约为(165±5)CFU/m L,高于国标限值(100 CFU/m L)。通过计算得出,检测600 m L本实验水样中的细菌菌落总数相当于检测1 000 m L(1 L)100 CFU/m L的自来水。首先将600 m L及低于和高于该体积不同体积梯度的自来水经滤膜过滤收集菌体;然后使用细菌基因组提取试剂盒提取滤膜上的细菌宏基因组DNA,电泳分析提取的宏基因组质量;最后检测各个体积自来水宏基因组DNA浓度,根据检测结果分析自来水中宏基因组DNA含量与细菌菌落总数之间的相关性。提取的自来水宏基因组DNA主条带清晰,可以进行基因组DNA含量测定分析;自来水中细菌宏基因组DNA浓度与细菌菌落总数之间呈正相关的线性关系。建立一种通过检测自来水中细菌宏基因组DNA浓度来反映细菌菌落总数的检测方法,检测灵敏度可达7 CFU/m L,且省时、重复性好。     

基于邻域粗糙集的连续值分布式数据属性约简

为了去除系统中的冗余属性,保持系统的分类能力,研究了连续值分布式数据的属性约简.给出了连续值分布式决策信息系统中邻域粗糙集的定义,讨论了分布式连续值决策信息系统中正域计算的可分解性.以保持分布式决策信息系统的正域不变为前提,探讨了分布式决策信息系统中属性的可约性,提出了分布式连续值决策信息系统的属性约简算法.为了验证该算法的有效性,在7份数据集上进行了3组实验.实验使用提出的算法对分布式数据进行属性约简,进而采用加权集成的方式进行分类测试.实验结果表明,该算法能够有效去除连续值分布式数据中的冗余属性,使得约简后的连续值分布式数据的集成分类能力与约简前相差不大.甚至更高.     

粗糙集文本分类的应用研究

文章提出了一种利用粗糙集理论生成文本分类规则的方法.首先,抽取特征词并计算权重.然后,在权值离散化之后,构造决策表.其中,特征词作为条件属性,类别作为决策属性.之后,将文本用属性约简和属性相对约简进行处理,得出决策规则.最后给出分类算法.     

一种基于区分矩阵的实值属性约简算法

通过粗糙集理论对一种实值属性约简算法进行了研究,给出了实值决策系统属性约简的算法,并采用UCI中的数据集进行分析,实验结果表明:该约简方法可以选择较少的属性而保持或改善分类能力.     

基于粗糙集的不相容决策表属性约简算法

对Skowron可辨识矩阵方法进行分析,应用反例说明基于Skowron可辨识矩阵方法对不相容决策表属性约简中存在一定的局限性.针对这一问题,提出了一种基于互信息的求属性核方法,并在此基础上利用互信息作为启发信息,在算法中加入了消除冗余属性的二次约简过程,构造一种完备的启发式属性约简算法.实例分析表明该算法能够有效地对不相容决策表进行属性约简,且具有较好的约简效果.     

基于粗糙集属性变分区的属性约简

应用粗糙集的方法,分析决策系统中不同的属性分类方法,以及不同分类方法引起的属性重要性与属性相对约简极小子集的变化情况,寻求属性分类方法与属性约简结果相互影响的内在因素,给出高效的属性分类方法和合理确定约简子集的策略,生成策略对应软件的实现算法,并运用软件实现算法来选取相对约简子集．试验结果显示了该策略及算法的有效性．     

金针菇rDNA-ITS序列分析

为研究金针菇rDNA-ITS序列特征,提取了不同来源金针菇子实体的基因组DNA,对其rDNA的内转录间隔区进行PCR扩增和序列测定,并提交NCBI获得登录号;将不同产地的金针菇ITS序列与本研究获得的金针菇ITS序列进行对比分析,利用ITS序列分析技术研究其遗传多样性,并构建了NJ系统发育树。结果表明,所有供试材料的ITS全长在710～719　bp范围内,G+C含量在49.3　%～49.9　%,所有序列共有34个变异位点,16个信息位点,ITS区遗传距离变化范围为0～0.016,即不同产地的金针菇ITS序列在进化过程中存在差异。此结果可为真菌分类鉴定等研究提供重要的资料和依据。     

Reagents for the site-specific cleavage of megabase DNA.

The physical mapping of chromosomes will be facilitated by methods of breaking large DNA into manageable fragments, or cutting uniquely at genetic markers of interest. Key issues in the design of sequence-specific DNA cleaving reagents are the specificity of binding, the number of different sequences that can be targeted and the cleavage yield.     

DNA序列数据的聚类挖掘

玻译基因图是二十一世纪最重要的任务之一．人类基因草图绘制后,接下来需要寻找各种基因的精确位置,研究各种基因的功能等．为此,必须对DNA序列片段进行分类,然后归类研究。本文分别从数学、物理、化学三个角度讨论了40个人工DNA序列和182个自然DNA序列的分类问题,提出了一些设想。     

DNA序列特征提取方法研究

针对DNA序列分类问题提出了两种特征提取方法,利用可分支持向量分类机间隔大、推广能力强的原理建立了DNA序列特征提取方法优劣的评价标准,利用该标准把本文的两种特征提取方法进行了比较,且跟以往的DNA序列特征提取方法进行了比较.实验表明,提出的两种特征方法得到的DNA序列特征完全能够代表DNA序列,对已知分类样本的预测率为100%,且此特征提取方法有很强的推广能力.     

Amplification of specific DNA sequences correlates with multi-drug resistance in Chinese hamster cells

Mammalian cells selected for resistance to certain cytotoxic drugs frequently develop cross-resistance to a broad spectrum of other drugs unrelated in structure to the original selective agent. This phenomenon constitutes a major problem in cancer chemotherapy. Multi-drug resistance arises from decreased intracellular drug accumulation, apparently due to an alteration of the plasma membrane. The observation of double minute chromosomes or homogeneously staining regions in some of the multi-drug-resistant cell lines suggests that gene amplification underlies this phenomenon. We have used the technique of DNA renaturation in agarose gels to detect, compare and clone amplified DNA sequences in Adriamycin- and colchicine-resistant sublines of Chinese hamster cells. We show that both Adriamycin- and colchicine-resistant cells contain amplified DNA fragments, some of which are amplified in both of these independently derived cell lines. Furthermore, loss of the multi-drug resistance phenotype on growth in the absence of drugs correlates with the loss of amplified DNA. These results strongly suggest that the DNA sequences which are amplified in common in multi-drug-resistant cell lines include the gene(s) responsible for a common mechanism of multi-drug resistance in these cells. We have cloned one of the commonly amplified DNA fragments and show that the degree of amplification of this fragment in the cells correlates with the degree of their drug resistance.     

A general method for site-directed mutagenesis in prokaryotes

The genetic analysis of genes from prokaryotic species for which experimental genetic systems have not yet been developed is often limited by the difficulty of producing mutations in those genes. We report here a general technique applicable to Gram-negative prokaryotes for site-directed mutagenesis of cloned DNA fragments which we have applied to the study of the symbiotic nitrogen fixation genes of Rhizobium meliloti. In particular, we mutagenized cloned R. meliloti restriction fragments in Escherichia coli with transposon Tn5 and then replaced the wild-type parental DNA sequences with the mutant DNA sequences in the R. meliloti genome. Using this method we show that an R. meliloti DNA restriction fragment, cloned previously on the basis of homology to Klebsiella pneumoniae nif genes, contains gene(s) essential for symbiotic nitrogen fixation. In addition, we use this method to construct a physical genetic map of a subset of the R. meliloti nif genes.     

Structure of a family of rat amylase genes

The sequences of two cloned rat pancreatic amylase cDNAs comprising 95% of the mRNA sequence are reported. Analysis of cloned rat genomic DNA fragments using cloned cDNA probes indicates that the rat genome contains multiple closely related amylase genes in which the cDNA sequences are distributed within a region 9 kilobases in length and are interrupted by at least seven intervening sequences.     

Genetic mapping of telomere-associated sequences in soybean ( Glycine max )

Two telomere-associated sequences (TAS), named STAS8 and STAS10, were cloned from soybean genomic DNA using polymerase chain reaction (PCR) amplification. Southern analysis showed that they were sequences with moderate copy number in soybean genome. Sequence analysis demonstrated that STAS10 had tandemly arrayed con sensus sequences of TTTAGGG and TIAGGG . The mapping of these two TAS was performed with a population of F8 re combinant inbred line using restriction fragment length polymorphisms(RFLP). Seven out of nine polymorphic fragments were mapped to the most distal position of five linkage groups, Dla, F, G2, H and Q of soybean, and the other two loci were closely linked and mapped to two interstitial positions within linkage group D1a. The mapping of TAS in soybean is essential for completeness of a molecular genetic map of soybean.     

Amino-terminal fragments of Escherichia coli lac repressor bind to DNA.

The N-terminal fragments (residues 1-51 and 1-59) obtained by selective tryptic cleavage of native lac repressor retain the ability to bind DNA. These fragments (headpieces) are monomeric and form complexes which resemble those of tetrameric repressor with non-operator DNA. But, they do not show the high specificity of repressor for operator sequences. The DNA binding has been demonstrated by filter-binding assay as well as in solution using absorption, circular dichroism, and fluorescence measurements.