共找到 20 条结果
In the early 1990s, after more than three decades of studying algorithms within the frame work of theoretical computer science, I shifted my focus to alogrithmic problems arising in genomics. There is a fundamental difference between the views of algorithms in the two fields: in theoretical computer science the input-output behavior of an algorithm is rigorously specified in advance, whereas in computational biology an algorithm is merely a vehicle for discovering Nature's ground truth. In order to be effective in computational genomics I have had to radically change my approach to research. On the occasion of this keynote address I will share some of the lessons I have learned, in the hope of making the way easier for computer scientists and mathematicians entering this field. These lessons will be encapsulated in a list of aphorisms, accompanied by illustrative examples.
Peptide-based vaccines, in which small peptides derived from target proteins (eptiopes) are used to provoke an immune reaction, have attracted considerable attention recently as a potential means both of treating infectious diseases and promoting the destruction of cancerous cells by a patient's own immune system. With the availability of large sequence databases and computers fast enough for rapid processing of large numbers of peptides, computer aided design of peptide-based vaccines has emerged as a promising approach to screening among billions of possible immune-active peptides to find those likely to provoke an immune response to a particular cell type. In this paper, we describe the development of three novel classes of methods for the prediction problem. We present a quadratic programming approach that can be trained on quantitative as well as qualitative data. The second method uses linear programming to counteract the fact that our training data contains mostly positive examples. The third class of methods uses sequence profiles obtained by clustering known epitopes to score candidate peptides. By integrating these methods, using a simple voting heuristic, we achieve improved accuracy over the state of the art.
With the development of microarray techniques, there is an increasing need of information processing methods to analyze the high throughput data. Clustering is one of the most promising candidates because of its simplicity, flexibility and robustness. However, there is no "perfect" clustering approach outperforming its counterparts, and it is hard to evaluate and combine the results from different techniques, especially in a field without much prior knowledge, such as bioinformatics. This paper proposes a meta-clustering approach to extract the information from results of different clustering techniques, so that a better interpretation of the data distribution can be obtained. A special distance measure is defined to represent the statistical "signal" of each cluster produced by various clustering techniques. The algorithm is applied on both artificial and real data Simulations show that the proposed approach is able to extract the information efficiently and accurately from the input clustering structure.
Analyzing protein sequence data becomes increasingly important recently. Most previous work on this area has mainly focused on building classification models. In this paper, we investigate in the problem of automatic clustering of unlabeled protein sequences. As a widely recognized technique in statistics and computer science, clustering has been proven very useful in detecting unknown object categories and revealing hidden correlations among objects. One difficulty that prevents clustering from being performed directly on protein sequence is the lack of an effective similarity measure that can be computed efficiently. Therefore, we propose a novel model for protein sequence cluster by exploring significant statistical properties possessed by the sequences. The concept of imprecise probabilities are introduced to the original probabilistic suffix tree to monitor the convergence of the empirical measurement and to guide the clustering process. It has been demonstrated that the proposed method can successfully discover meaningful families without the necessity of learning models of different families from pre-labeled "training data".
The personal information encoded in genomic DNA should not be made available to the public. With the increasing discoveries of new genes, it has become necessary to establish a security system for personal genome information. Although many security systems that are applied for electrical information in computers have been developed and established, there is no security system for information at DNA level. In this paper, we describe a new security system for information encoded within DNA. The original genomic DNA was mixed with many kinds of dummy DNAs (mixtures of natural and/or artificial DNAs) resulting in the masking of the original information. Using these dummy molecules, we succeeded to completely 'lock'the original genome information. If this information must be 'unlocked', it can be extracted and analyzed by a removal of dummy DNAs using molecular tagging techniques or by selective amplification using key primers.
Comparative analysis of syntenic genome sequences can be used to identify functional sites such as exons and regulatory elements. Here, the first step is to align two or several evolutionary related sequences and, in recent years, a number of computer programs have been developed for alignment of large genomic sequences. Some of these programs are extremely fast but often time-efficiency is achieved at the expense of sensitivity. One way of combining speed and sensitivity is to use an anchored-alignment approach. In a first step, a fast heuristic identifies a chain of strong sequence similarities that serve as anchor points. In a second step, regions between these anchor points are aligned using a slower but more sensitive method. We present CHAOS, a novel algorithm for rapid identification of chains of local sequence similarities among large genomic sequences. Similarities identified by CHAOS are used as anchor points to improve the running time of the DIALIGN alignment program. Systematic test runs show that this method can reduce the running time of DIALIGN by more than 93% while affecting the quality of the resulting alignments by only 1%. The source code for CHAOS is available at http://www.stanford.edu/~brudno/chaos/ An integrated program package containing CHAOS and DIALIGN is available at http://bibiserv.techfak.uni-bielefeld.de/dialign/
Bioinformatics has become an active research area in recent years. The amount of mapped sequences doubles every fourteen months. BLAST has been widely employed for retrieving sequences which has similar portion(s) to a given sequence. However, BLAST has to scan the entire database every time when a query is issued. This can be very time consuming especially when the database is large. In this paper, we study the problem on how to build a persistent index structure for protein sequences to support approximate match. The suffix tree has been proposed as a solution to index sequence database and has been deployed on organizing DNA sequences (Hunt et al. 2001). Unfortunately, it suffers from the problem of "memory bottleneck" that prevents it from being applied efficiently to a large database. The performance even degrades further for protein database due to a larger fanout at each node. Here, we employ an indexing structure, called BASS-tree, to support approximate match in sublinear time on a large protein database. We call this indexing method as sequence approximate match (SAM) index method. The search of approximate matches can be properly directed to the portion in the database with a high potential of matching quickly. It has been demonstrated in our experiments that the potential performance improvement is in an order of magnitude over alternative methods such as the BLAST algorithm and the suffix tree.
The Cray MTA-2 (Multithreaded Architecture) is an unusual parallel supercomputer that promises ease of use and high performance. We describe our experience on the MTA-2 with a molecular dynamics code, SIMU-MD, that we are using to simulate the translocation of DNA through a nanopore in a silicon based ultrafast sequencer. Our sequencer is constructed using standard VLSI technology and consists of a nanopore surrounded by Field Effect Transistors (FETs). We propose to use the FETs to sense variations in charge as a DNA molecule translocates through the pore and thus differentiate between the four building block nucleotides of DNA. We were able to port SIMU-MD, a serial C code, to the MTA with only a modest effort and with good performance. Our porting process needed neither a parallelism support platform nor attention to the intimate details of parallel programming and interprocessor communication, as would have been the case with more conventional supercomputers.
We have constructed a computer model of the cytotoxic T lymphocyte (CTL) response to antigen and the maintenance of immunological memory. Because immune responses often begin with small numbers of cells and there is great variation among individual immune systems, we have chosen to implement a stochastic model that captures the life cycle of T cells more faithfully than deterministic models. Past models of the immune response have been differential equation based, which do not capture stochastic effects, or agent-based, which are computationally expensive. We use a stochastic stage-structured approach that has many of the advantages of agent-based modeling but is much more efficient. Our model can provide insights into the effect infections have on the CTL repertoire and the response to subsequent infections.
NMR resonance assignment is one of the key steps in solving an NMR protein structure. The assignment process links resonance peaks to individual residues of the target protein sequence, providing the prerequisite for establishing intra- and inter-residue spatial relationships between atoms. The assignment process is tedious and time-consuming, which could take many weeks. Though there exist a number of computer programs to assist the assignment process, many NMR labs are still doing the assignments manually to ensure quality. This paper presents a new computational method based on our recent work towards automating the assignment process, particularly the process of backbone resonance peak assignment. We formulate the assignment problem as a constrained weighted bipartite matching problem. While the problem, in the most general situation, is NP-hard, we present an efficient solution based on a branch-and-bound algorithm with effective bounding techniques and a greedy filtering algorithm for reducing the search space. Our experimental results on 70 instances of (pseudo) real NMR data derived from 14 proteins demonstrate that the new solution runs much faster than a recently introduced (exhaustive) two-layer algorithm and recovers more correct peak assignments than the two-layer algorithm.
NMR resonance peak assignment is one of the key steps in solving an NMR protein structure. The assignment process links resonance peaks to individual residues of the target protein sequence, providing the prerequisite for establishing intra- and inter-residue spatial relationships between atoms. The assignment process is tedious and time-consuming, which could take many weeks. Though there exist a number of computer programs to assist the assignment process, many NMR labs are still doing the assignments manually to ensure quality. This paper presents (1) a new scoring system for mapping spin systems to residues, (2) an automated adjacency information extraction procedure from NMR spectra, and (3) a very fast assignment algorithm based on our previous proposed greedy filtering method and a maximum matching algorithm to automate the assignment process. The computational tests on 70 instances of (pseudo) experimental NMR data of 14 proteins demonstrate that the new score scheme has much better discerning power with the aid of adjacency information between spin systems simulated across various NMR spectra. Typically, with automated extraction of adjacency information, our method achieves nearly complete assignments for most of the proteins. The experiment shows very promising perspective that the fast automated assignment algorithm together with the new score scheme and automated adjacency extraction may be ready for practical use.
This paper presents a novel algorithm for identification and functional characterization of "key" genome features responsible for a particular biochemical process of interest. The central idea is that individual genome features are identified as "key" features if the discrimination accuracy between two classes of genomes with respect to a given biochemical process is sufficiently affected by the inclusion or exclusion of these features. In this paper, genome features are defined by high-resolution gene functions. The discrimination procedure utilizes the Support Vector Machine classification technique. The application to the oxygenic photosynthetic process resulted in 126 highly confident candidate genome features. While many of these features are well-known components in the oxygenic photosynthetic process, others are completely unknown, even including some hypothetical proteins. It is obvious that our algorithm is capable of discovering features related to a targeted biochemical process.
The cancer state of a cell is characterized by alterations of important cellular processes such as cell proliferation, apoptosis, DNA-damage repair, etc. The expression of genes associated with cancer related pathways, therefore, may exhibit differences between the normal and the cancerous states. We explore various means to find these differences. We analyze 6 different pathways (p53, Ras, Brca, DNA damage repair, NFkappab and beta-catenin) and 4 different types of cancer: colon, pancreas, prostate and kidney. Our results are found to be mostly consistent with existing knowledge of the involvement of these pathways in different cancers. Our analysis constitutes proof of principle that it may be possible to predict the involvement of a particular pathway in cancer or other diseases by using gene expression data. Such method would be particularly useful for the types of diseases where biology is poorly understood.
DNA microarray technology provides a broad snapshot of the state of the cell by measuring the expression levels of thousands of genes simultaneously. Visualization techniques can enable the exploration and detection of patterns and relationships in a complex dataset by presenting the data in a graphical format in which the key characteristics become more apparent. The purpose of this study is to present an interactive visualization technique conveying the temporal patterns of gene expression data in a form intuitive for non-specialized end-users. The first Fourier harmonic projection (FFHP) was introduced to translate the multi-dimensional time series data into a two dimensional scatter plot. The spatial relationship of the points reflect the structure of the original dataset and relationships among clusters become two dimensional. The proposed method was tested using two published, array-derived gene expression datasets. Our results demonstrate the effectiveness of the approach.
Oligonucleotide fingerprinting is a powerful DNA array based method to characterize cDNA and ribosomal RNA gene (rDNA) libraries and has many applications including gene expression profiling and DNA clone classification. We are especially interested in the latter application. A key step in the method is the cluster analysis of fingerprint data obtained from DNA array hybridization experiments. Most of the existing approaches to clustering use (normalized) real intensity values and thus do not treat positive and negative hybridization signals equally (positive signals are much more emphasized). In this paper, we consider a discrete approach. Fingerprint data are first normalized and binarized using control DNA clones. Because there may exist unresolved (or missing) values in this binarization process, we formulate the clustering of (binary) oligonucleotide fingerprints as a combinatorial optimization problem that attempts to identify clusters and resolve the missing values in the fingerprints simultaneously. We study the computational complexity of this clustering problem and a natural parameterized version, and present an efficient greedy algorithm based on MINIMUM CLIQUE PARTITION on graphs. The algorithm takes advantage of some unique properties of the graphs considered here, which allow us to efficiently find the maximum cliques as well as some special maximal cliques. Our experimental results on simulated and real data demonstrate that the algorithm runs faster and performs better than some popular hierarchical and graph-based clustering methods. The results on real data from DNA clone classification also suggest that this discrete approach is more accurate than clustering methods based on real intensity values, in terms of separating clones that have different characteristics with respect to the given oligonucleotide probes.
We propose a statistical method for estimating a gene network based on Bayesian networks from microarray gene expression data together with biological knowledge including protein-protein interactions, protein-DNA interactions, binding site information, existing literature and so on. Unfortunately, microarray data do not contain enough information for constructing gene networks accurately in many cases. Our method adds biological knowledge to the estimation method of gene networks under a Bayesian statistical framework, and also controls the trade-off between microarray information and biological knowledge automatically. We conduct Monte Carlo simulations to show the effectiveness of the proposed method. We analyze Saccharomyces cerevisiae gene expression data as an application.
The cWINNOWER algorithm detects fuzzy motifs in DNA sequences rich in protein-binding signals. A signal is defined as any short nucleotide pattern having up to d mutations differing from a motif of length l. The algorithm finds such motifs if multiple mutated copies of the motif (i.e., the signals) are present in the DNA sequence in sufficient abundance. The cWINNOWER algorithm substantially improves the sensitivity of the winnower method of Pevzner and Sze by imposing a consensus constraint, enabling it to detect much weaker signals. We studied the minimum number of detectable motifs qc as a function of sequence length N for random sequences. We found that q(c) increases linearly with N for a fast version of the algorithm based on counting three-member sub-cliques. Imposing consensus constraints reduces q(c) by a factor of three in this case, which makes the algorithm dramatically more sensitive. Our most sensitive algorithm, which counts four-member sub-cliques, needs a minimum of only 13 signals to detect motifs in a sequence of length N = 12,000 for (l,d) = (15,4).
We have developed an algorithm called the Universal Chemical Key (UCK) algorithm that constructs a unique key for a molecular structure. The molecular structures are represented as undirected labeled graphs with the atoms representing the vertices of the graph and the bonds representing the edges. The algorithm was tested on 236,917 compounds obtained from the National Cancer Institute (NCI) database of chemical compounds. In this paper we present the algorithm,some examples and the experimental results on the NCI database. On the NCI database, the UCK algorithm provided distinct unique keys for chemicals with different molecular structures.