Discriminant analysis and its application in DNA sequence motif recognition

Identification of functional motifs in a DNA sequence is fundamentally a statistical pattern recognition problem. Discriminant analysis is widely used for solving such problems. This paper will review two basic parametric methods: LDA (linear discriminant analysis) and QDA (quadratic discriminant analysis). Their usage in recognition of splice sites and exons in the human genome will be demonstrated.     

A fractal method to distinguish coding and non-coding sequences in a complete genome based on a number sequence representation

A fractal method to distinguish coding and non-coding sequences in a complete genome is proposed, based on different statistical behaviors between these two kinds of sequences. We first propose a number sequence representation of DNA sequences. Multifractal analysis is then performed on the measure representation of the obtained number sequence. The three exponents C(-1), C1 and C2 are selected from the result of multifractal analysis. Each DNA may be represented by a point in the three-dimensional space generated by these three-component vectors. It is shown that points corresponding to coding and non-coding sequences in the complete genome of many prokaryotes are roughly distributed in different regions. Fisher's discriminant algorithm can be used to separate these two regions in the spanned space. If the point (C(-1),C1,C2) for a DNA sequence is situated in the region corresponding to coding sequences, the sequence is discriminated as a coding sequence; otherwise, the sequence is classified as a non-coding one. For all 51 prokaryotes we considered , the average discriminant accuracies pc,pnc,qc and qnc reach 72.28%, 84.65%, 72.53% and 84.18%, respectively.     

In silico pattern-based analysis of the human cytomegalovirus genome

More than 200 open reading frames (ORFs) from the human cytomegalovirus genome have been reported as potentially coding for proteins. We have used two pattern-based in silico approaches to analyze this set of putative viral genes. With the help of an objective annotation method that is based on the Bio-Dictionary, a comprehensive collection of amino acid patterns that describes the currently known natural sequence space of proteins, we have reannotated all of the previously reported putative genes of the human cytomegalovirus. Also, with the help of MUSCA, a pattern-based multiple sequence alignment algorithm, we have reexamined the original human cytomegalovirus gene family definitions. Our analysis of the genome shows that many of the coded proteins comprise amino acid combinations that are unique to either the human cytomegalovirus or the larger group of herpesviruses. We have confirmed that a surprisingly large portion of the analyzed ORFs encode membrane proteins, and we have discovered a significant number of previously uncharacterized proteins that are predicted to be G-protein-coupled receptor homologues. The analysis also indicates that many of the encoded proteins undergo posttranslational modifications such as hydroxylation, phosphorylation, and glycosylation. ORFs encoding proteins with similar functional behavior appear in neighboring regions of the human cytomegalovirus genome. All of the results of the present study can be found and interactively explored online (http://cbcsrv.watson.ibm.com/virus/).     

一种基于特征筛选的原核生物启动子判别分析方法

启动子识别是研究基因转录调控的重要环节,但目前方法的识别正确率偏低。在深入分析原核启动子特征的基础上,提出了一种基于特征筛选的原核启动子判别分析方法,首先在启动子序列的组成特征、信号特征和结构特征中选取备选特征,为每个特征建立适当的描述模型,并对主要的保守模式采用复合模式模型;再通过模型计算对备选特征进行逐步筛选,优化特征集,将序列表示为组合特征向量;最终利用二次判别分析实现识别。对大肠杆菌和枯草杆菌实际启动子数据进行的刀切法测试验证了方法的有效性和通用性。对于大肠杆菌非编码区(70启动子,识别的平均正确率达到了85.8%,优于其它几种典型识别方法;对于大肠杆菌编码区内部)70启动子和其它几种原核启动子,平均正确率也都超过了80%。方法框架还具有良好的可扩展性,能够方便地容纳新特征,使识别性能不断提高。     

Isochore chromosome maps of the human genome

The human genome is a mosaic of isochores, which are long DNA segments (300 kbp) relatively homogeneous in G+C. Human isochores were first identified by density-gradient ultracentrifugation of bulk DNA, and differ in important features, e.g. genes are found predominantly in the GC-richest isochores. Here, we use a reliable segmentation method to partition the longest contigs in the human genome draft sequence into long homogeneous genome regions (LHGRs), thereby revealing the isochore structure of the human genome. The advantages of the isochore maps presented here are: (1) sequence heterogeneities at different scales are shown in the same plot; (2) pair-wise compositional differences between adjacent regions are all statistically significant; (3) isochore boundaries are accurately defined to single base pair resolution; and (4) both gradual and abrupt isochore boundaries are simultaneously revealed. Taking advantage of the wide sample of genome sequence analyzed, we investigate the correspondence between LHGRs and true human isochores revealed through DNA centrifugation. LHGRs show many of the typical isochore features, mainly size distribution, G+C range, and proportions of the isochore classes. The relative density of genes, Alu and long interspersed nuclear element repeats and the different types of single nucleotide polymorphisms on LHGRs also coincide with expectations in true isochores. Potential applications of isochore maps range from the improvement of gene-finding algorithms to the prediction of linkage disequilibrium levels in association studies between marker genes and complex traits. The coordinates for the LHGRs identified in all the contigs longer than 2 Mb in the human genome sequence are available at the online resource on isochore mapping: http://bioinfo2.ugr.es/isochores.     

Isochore chromosome maps of eukaryotic genomes

Analytical DNA ultracentrifugation revealed that eukaryotic genomes are mosaics of isochores: long DNA segments (>300 kb on average) relatively homogeneous in G+C. Important genome features are dependent on this isochore structure, e.g. genes are found predominantly in the GC-richest isochore classes. However, no reliable method is available to rigorously partition the genome sequence into relatively homogeneous regions of different composition, thereby revealing the isochore structure of chromosomes at the sequence level. Homogeneous regions are currently ascertained by plain statistics on moving windows of arbitrary length, or simply by eye on G+C plots. On the contrary, the entropic segmentation method is able to divide a DNA sequence into relatively homogeneous, statistically significant domains. An early version of this algorithm only produced domains having an average length far below the typical isochore size. Here we show that an improved segmentation method, specifically intended to determine the most statistically significant partition of the sequence at each scale, is able to identify the boundaries between long homogeneous genome regions displaying the typical features of isochores. The algorithm precisely locates classes II and III of the human major histocompatibility complex region, two well-characterized isochores at the sequence level, the boundary between them being the first isochore boundary experimentally characterized at the sequence level. The analysis is then extended to a collection of human large contigs. The relatively homogeneous regions we find show many of the features (G+C range, relative proportion of isochore classes, size distribution, and relationship with gene density) of the isochores identified through DNA centrifugation. Isochore chromosome maps, with many potential applications in genomics, are then drawn for all the completely sequenced eukaryotic genomes available.     

A novel approach for identifying candidate imprinted genes through sequence analysis of imprinted and control genes

Through the sequence analysis of 27 imprinted human genes and a set of 100 control genes we have developed a novel approach for identifying candidate imprinted genes based on the differences in sequence composition observed. The imprinted genes were found to be associated with significantly reduced numbers of short interspersed transposable element (SINE) Alus and mammalian-wide interspersed repeat (MIR) repeat elements, as previously reported. In addition, a significant association between imprinted genes and increased numbers of low-complexity repeats was also evident. Numbers of the Alu classes AluJ and AluS were found to be significantly depleted in some parts of the flanking regions of imprinted genes. A recent study has proposed that there is active selection against SINE elements in imprinted regions. Alternatively, there may be differences in the rates of insertion of Alu elements. Our study indicates that this difference extends both upstream and downstream of the coding region. This and other consistent differences between the sequence characteristics of imprinted and control genes has enabled us to develop discriminant analysis, which can be used to screen the genome for candidate imprinted genes. We have applied this function to a number of genes whose imprinting status is disputed or uncertain.     

The PEDANT genome database

The PEDANT genome database (http://pedant.gsf.de) provides exhaustive automatic analysis of genomic sequences by a large variety of established bioinformatics tools through a comprehensive Web-based user interface. One hundred and seventy seven completely sequenced and unfinished genomes have been processed so far, including large eukaryotic genomes (mouse, human) published recently. In this contribution, we describe the current status of the PEDANT database and novel analytical features added to the PEDANT server in 2002. Those include: (i) integration with the BioRS data retrieval system which allows fast text queries, (ii) pre-computed sequence clusters in each complete genome, (iii) a comprehensive set of tools for genome comparison, including genome comparison tables and protein function prediction based on genomic context, and (iv) computation and visualization of protein-protein interaction (PPI) networks based on experimental data. The availability of functional and structural predictions for 650 000 genomic proteins in well organized form makes PEDANT a useful resource for both functional and structural genomics.     

Long-range autocorrelations of CpG islands in the human genome

In this paper, we use a statistical estimator developed in astrophysics to study the distribution and organization of features of the human genome. Using the human reference sequence we quantify the global distribution of CpG islands (CGI) in each chromosome and demonstrate that the organization of the CGI across a chromosome is non-random, exhibits surprisingly long range correlations (10 Mb) and varies significantly among chromosomes. These correlations of CGI summarize functional properties of the genome that are not captured when considering variation in any particular separate (and local) feature. The demonstration of the proposed methods to quantify the organization of CGI in the human genome forms the basis of future studies. The most illuminating of these will assess the potential impact on phenotypic variation of inter-individual variation in the organization of the functional features of the genome within and among chromosomes, and among individuals for particular chromosomes.     

HCV genotyping using statistical classification approach

The genotype of Hepatitis C Virus (HCV) strains is an important determinant of the severity and aggressiveness of liver infection as well as patient response to antiviral therapy. Fast and accurate determination of viral genotype could provide direction in the clinical management of patients with chronic HCV infections. Using publicly available HCV nucleotide sequences, we built a global Position Weight Matrix (PWM) for the HCV genome. Based on the PWM, a set of genotype specific nucleotide sequence "signatures" were selected from the 5' NCR, CORE, E1, and NS5B regions of the HCV genome. We evaluated the predictive power of these signatures for predicting the most common HCV genotypes and subtypes. We observed that nucleotide sequence signatures selected from NS5B and E1 regions generally demonstrated stronger discriminant power in differentiating major HCV genotypes and subtypes than that from 5' NCR and CORE regions. Two discriminant methods were used to build predictive models. Through 10 fold cross validation, over 99% prediction accuracy was achieved using both support vector machine (SVM) and random forest based classification methods in a dataset of 1134 sequences for NS5B and 947 sequences for E1. Prediction accuracy for each genotype is also reported.     

Lineage-Specific Biology Revealed by a Finished Genome Assembly of the Mouse

The mouse (Mus musculus) is the premier animal model for understanding human disease and development. Here we show that a comprehensive understanding of mouse biology is only possible with the availability of a finished, high-quality genome assembly. The finished clone-based assembly of the mouse strain C57BL/6J reported here has over 175,000 fewer gaps and over 139 Mb more of novel sequence, compared with the earlier MGSCv3 draft genome assembly. In a comprehensive analysis of this revised genome sequence, we are now able to define 20,210 protein-coding genes, over a thousand more than predicted in the human genome (19,042 genes). In addition, we identified 439 long, non–protein-coding RNAs with evidence for transcribed orthologs in human. We analyzed the complex and repetitive landscape of 267 Mb of sequence that was missing or misassembled in the previously published assembly, and we provide insights into the reasons for its resistance to sequencing and assembly by whole-genome shotgun approaches. Duplicated regions within newly assembled sequence tend to be of more recent ancestry than duplicates in the published draft, correcting our initial understanding of recent evolution on the mouse lineage. These duplicates appear to be largely composed of sequence regions containing transposable elements and duplicated protein-coding genes; of these, some may be fixed in the mouse population, but at least 40% of segmentally duplicated sequences are copy number variable even among laboratory mouse strains. Mouse lineage-specific regions contain 3,767 genes drawn mainly from rapidly-changing gene families associated with reproductive functions. The finished mouse genome assembly, therefore, greatly improves our understanding of rodent-specific biology and allows the delineation of ancestral biological functions that are shared with human from derived functions that are not.     

Prediction of operator-binding protein by discriminant analysis

A number of operator-binding proteins contain similar sequence features to Cro and cI repressors of bacteriophage and CAP protein of Escherichia coli, such as conserved amino acids at constant positions. However, these sequence patterns also occur in proteins that are not operator-binding. We use sequence analogy information in conjunction with a pattern recognition algorithm. The functional and structural properties, e.g., distributions of hydrophobicity, hydrophilicity, charged amino acids, electrostatic free energy, and helical structures of protein are also considered. Within the framework of discriminant analysis, we calculate the above variables and search for a better combination of variables. To assess the discriminatory power of these variables, we allocated additional sequences and predict DNA-binding regions of regulatory proteins not included in the training set.     

Identification of a conserved sequence in the non-coding regions of many human genes.

We have analyzed a sequence of approximately 70 base pairs (bp) that shows a high degree of similarity to sequences present in the non-coding regions of a number of human and other mammalian genes. The sequence was discovered in a fragment of human genomic DNA adjacent to an integrated hepatitis B virus genome in cells derived from human hepatocellular carcinoma tissue. When one of the viral flanking sequences was compared to nucleotide sequences in GenBank, more than thirty human genes were identified that contained a similar sequence in their non-coding regions. The sequence element was usually found once or twice in a gene, either in an intron or in the 5' or 3' flanking regions. It did not share any similarities with known short interspersed nucleotide elements (SINEs) or presently known gene regulatory elements. This element was highly conserved at the same position within the corresponding human and mouse genes for myoglobin and N-myc, indicating evolutionary conservation and possible functional importance. Preliminary DNase I footprinting data suggested that the element or its adjacent sequences may bind nuclear factors to generate specific DNase I hypersensitive sites. The size, structure, and evolutionary conservation of this sequence indicates that it is distinct from other types of short interspersed repetitive elements. It is possible that the element may have a cis-acting functional role in the genome.