siDirect 2.0 updated software for designing functional siRNA with reduced seed%2Ddependent off%2Dtarget effect.

UTGB toolkit for personalized genome browsers.

MachiBase a Drosophila melanogaster 5%27%2Dend mRNA transcription database.

Relational%2Dstyle XML query.

UTGB%2Fmedaka genomic resource database for medaka biology.

Efficient Integration of Structure Indexes of XML.

Amoeba Join Overcoming Structural Fluctuations in XML Data.
There are no universal rules for organizing data in XML. Consequently, semantically identical XML documents may have different structures; we call this structural fluctuation in XML. Finding all the structural fluctuations in an XML document requires verbose path expression queries. To overcome this problem, we developed a novel query processing primitive, called amoeba join. Amoeba join does not require explicit path structures in query statements; tag names or keywords are sufficient to perform searches. This paper introduces several amoeba join processing algorithms and demonstrates their performance. 1.
PrimerStation a highly specific multiplex genomic PCR primer design server for the human genome.

Accelerated off%2Dtarget search algorithm for siRNA.

5%27SAGE 5%27%2Dend Serial Analysis of Gene Expression database.

Data mining tools for the Saccharomyces cerevisiae morphological database.

dsCheck highly sensitive off%2Dtarget search software for double%2Dstranded RNA%2Dmediated RNA interference.

Data mining tools for the Saccharomyces cerevisiae morphological database.

Proceedings of the 4th ACM SIGKDD Workshop on Data Mining in Bioinformatics %28BIOKDD 2004%29

Itemset Classified Clustering.
The two most commonly addressed data mining tasks are  predictive modelling and clustering. Here we address the task of predictive  clustering, which contains elements of both and generalizes them  to some extent. We propose a novel approach to predictive clustering  called predictive clustering rules, present an initial implementation and  its preliminary experimental evaluation.
Constrained clusters of gene expression profiles with pathological features.
Motivation: Gene expression profiles should be useful in distinguishing variations in disease, since they reflect accurately the status of cells. The primary clustering of gene expression reveals the genotypes that are responsible for the proximity of members within each cluster, while further clustering elucidates the pathological features of the individual members of each cluster. However, since the first clustering process and the second classification step, in which the features are associated with clusters, are performed independently, the initial set of clusters may omit genes that are associated with pathologically meaningful features. Therefore, it is important to devise a way of identifying gene expression clusters that are associated with pathological features. Results: We present the novel technique of “itemset constrained clustering ”(IC-Clustering), which computes the optimal cluster that maximises the interclass variance of gene expression between groups, which are divided according to the restriction that only divisions that can be expressed using common features are allowed. This constraint automatically labels each cluster with a set of pathological features that characterize that cluster. When applied to liver cancer datasets, IC-Clustering revealed informative gene expression clusters, which could be annotated with various pathological features, such as “tumor ” and “man”, or “except tumor ” and “normal liver function. ” In contrast, the k-means method overlooked these clusters. Supplementary Information: Our dataset is available on the following web page:
Development of Image Processing Program for Yeast Cell Morphology.

Computing Highly Specific and Noise%2Dtolerant Oligomers Efficiently.

Computing Highly Specific and Noise%2Dtolerant Oligomers Efficiently.



siDirect highly effective
siDirect
Report on BIOKDD04 workshop on data mining in Bioinformatics.



A Fast and Sensitive Algorithm for Aligning Ests to the Human Genome.
Introduction  The Human Genome Project is an international collaboration, designed to investigate the genetic complexity of humans. Initially, the roughly three billion nucleotides of the human genome were elucidated (Celera Genomics    , International Human Genome Sequencing Consortium    ). The next step involves interpreting the encoded sequences. In order to identify the coding regions, i.e., regions containing exons and introns, of any given DNA sequence, many ESTs must be aligned with genomic DNA to reveal these complex structures, while verifying alternative splicing transcripts. The alignment of full-length cDNAs gives clues to regulatory elements in the upstream regions. Furthermore, the annotation of the upstream regions using Transfac data identifies the candidates of cis-elements. The alignment of both sequences associated with expression patterns and sequences from dbSNP with the identified locations of SNPs near the gene, enables more detailed functional 2 Jun Ogasawara a
Preface.
part of a series of studies in support of the California Governor’s Advisory Committee on Education Excellence. i PREFACE i LIST OF FIGURES iv
Practical Software for Aligning ESTs to Human Genome.

Fast and Sensitive Algorithm for Aligning ESTs to Human Genome.
Introduction  The Human Genome Project is an international collaboration, designed to investigate the genetic complexity of humans. Initially, the roughly three billion nucleotides of the human genome were elucidated (Celera Genomics    , International Human Genome Sequencing Consortium    ). The next step involves interpreting the encoded sequences. In order to identify the coding regions, i.e., regions containing exons and introns, of any given DNA sequence, many ESTs must be aligned with genomic DNA to reveal these complex structures, while verifying alternative splicing transcripts. The alignment of full-length cDNAs gives clues to regulatory elements in the upstream regions. Furthermore, the annotation of the upstream regions using Transfac data identifies the candidates of cis-elements. The alignment of both sequences associated with expression patterns and sequences from dbSNP with the identified locations of SNPs near the gene, enables more detailed functional 2 Jun Ogasawara a
Answering the Most Correlated N Association Rules Efficiently.
Many algorithms have been proposed for computing association  rules using the support-confidence framework. One drawback of  this framework is its weakness in expressing the notion of correlation.
Computing Optimal Hypotheses Efficiently for Boosting.
This paper sheds light on a strong connection between AdaBoost and several optimization algorithms for data mining. AdaBoost has been the subject of much interests as an effective methodology for classification task. AdaBoost repeatedly generates one hypothesis in each round, and finally it is able to make a highly accurate prediction by taking a weighted majority vote on the resulting hypotheses. Freund and Schapire have remarked that the use of simple hypotheses such as singletest decision trees instead of huge trees would be promising for achieving high accuracy and avoiding overfitting to the training data. One major drawback of this approach however is that accuracies of simple individual hypotheses may not always be high, hence demanding a way of computing more accurate (or, the most accurate) simple hypotheses efficiently.
The Gene Resource Locator gene locus maps for transcriptome analysis.

KDD Cup 2001 Report.
This paper presents results and lessons from KDD Cup 2001.
Data Mining with optimized two%2Ddimensional association rules.

Efficient Construction of Regression Trees with Range and Region Splitting.
We propose an efficient way of constructing regression trees in order to predict the objective numeric attribute values of given tuples. A regression tree is a rooted binary tree such that each internal node contains a test, which can be expressed as an RDB query, for splitting tuples into two disjoint classes and passing data in each class down to the left or right subtree. The mean of the objective attribute values at the leaf is used as the predicted value of the tuple.
BodyMap incorporated PCR%2Dbased expression profiling data and a gene ranking system.

Discovery Science
Introduction  Discovery has always attracted people and has been a major motiveforce  of progress in human history. It has long been discussed in the philosophyof  science but it is only after 1960&#039;s that the problem of discovery became properly addressed by new philosophy of science group  6)  . In the last two decades, there has also been a considerable progress, in the field of artificial intelligence, of a domain often considered the realm of genius - empirical discovery  5)  . Historically, discovery tends to be viewed as a difficult form of learning from observation although learning and discovery convey rather different meanings  4)  . The former suggests a gradual process, while the latter suggests a more rapid mental event, often involving some form of insight. Learning may lead to an unconscious change in knowledge, while one is always aware that a discovery has been made. The results of learning can be declarative or procedura
BodyMap a human and mouse gene expression database.
BodyMap is a human and mouse gene expression database that has been maintained since 1993. It is based on site-directed 3  -ESTs collected from nonbiased cDNA libraries constructed at Osaka University and contains &gt;270 000 sequences from 60 human and 38 mouse tissues. The site-directed nature of the sequence tags allows unequivocal grouping of tags representing the same transcript and provides abundance information for each transcript in different parts of the body. Our collection of ESTs was compared periodically with other public databases for cross referencing. The histological resolution of source tissues and unique cloning strategy that minimized cloning bias enabled BodyMap to support three unique mRNA based experiments in silico. First, the recurrence information for clones in each library provides a rough estimate of the mRNA composition of each source tissue. Second, a user can search the entire data set with nucleotide sequences or keywords to assess expression patterns of particular genes. Third, and most important, BodyMap allows a user to select genes that have a desired expression pattern in humans and mice. BodyMap is accessible through the WWW at http:// bodymap.ims.u-tokyo.ac.jp  
Towards Mining Gene Expression Database.

Weighted Majority Decision among Several Region Rules for Scientific Discovery.

Weighted Majority Decision among Region Rules for a Categorical Dataset.

Parallel Branch%2Dand%2DBound Graph Search for Correlated Association Rules.
There have been proposed efficient ways of enumerating all the association rules that are interesting with respect to support, confidence, or other measures. In contrast, we examine the optimization problem of computing the optimal association rule that maximizes the significance of the correlation between the assumption and the conclusion of the rule. We propose a parallel branch-and-bound graph search algorithm tailored to this problem. The key features of the design are (1) novel branchand -bound heuristics, and (2) a rule of rewriting conjunctions that avoids maintaining the list of visited nodes. Experiments on two different types of large-scale shared-memory multi-processors confirm that the speed-up of the computation time scales almost linearly with the number of processors, and the size of search space could be dramatically reduced by the branch-and-bound heuristics.  1 Introduction  Many organizations are seeking strategies for processing or interpreting massive amounts of da...
Mining Optimized Association Rules for Numeric Attributes.
Given a huge database, we address the problem of finding association rules for numeric attributes, such as (Balance c 1) + (CardLoan = yes), which implies that bank customers whose balances fall in a range 1 are likely to use card loan with a probability greater than p. The above rule is interesting only if the range 1 has some special feature with respect to the interrelation between Balance and CardLoan. It is required that the number of customers whose balances are contained in 1 (called the support of 1) is sufficient, and also that the probability p (called the confidence raho) should be much higher than the average probability of the condition being met. Our goal is to realize a system that finds such appropriate ranges automatically. We mainly focus on computing two
On Classification and Regression.
We address the problem of computing various types of expressive tests for decision tress and regression trees. Using expressive tests is promising, because it may improve the prediction accuracy of trees. The drawback is that computing an optimal test could be costly. We present a unified framework to approach this problem, and we revisit the design of efficient algorithms for computing important special cases. We also prove that it is intractable to compute an optimal conjunction or disjunction.
Computing Optimized Rectilinear Regions for Association Rules.

Efficient Construction of Regression Trees with Range and Region Splitting.
We propose an efficient way of constructing regression trees in order to predict the objective numeric attribute values of given tuples. A regression tree is a rooted binary tree such that each internal node contains a test, which can be expressed as an RDB query, for splitting tuples into two disjoint classes and passing data in each class down to the left or right subtree. The mean of the objective attribute values at the leaf is used as the predicted value of the tuple.
Interval Finding and Its Application to Data Mining.

Avoiding Cartesian products for multiple joins.
. Computing the natural join of a set of relations is an important operation in relational database systems. The ordering of joins determines to a large extent the computation time of the join. Since the number of possible orderings could be very large, query optimizers first reduce the search space by using various heuristics and then try to select an optimal ordering of joins. Avoiding Cartesian products is a common heuristic for reducing the search space, but it cannot guarantee optimal ordering in its search space, because the cheapest Cartesian-product-free (CPF, for short) ordering could be significantly worse than an optimal non-CPF ordering by a factor of an arbitrarily large number. In this paper, we use programs consisting of joins, semijoins, and projections for computing the join of some relations, and we introduce a novel algorithm that derives programs from CPF orderings of joins. We show that there exists a CPF ordering from which our algorithm derives a program whose co...
Interval Finding and Its Application to Data Mining.

Mining Optimized Association Rules for Numeric Attributes.
Given a huge database, we address the problem of finding association rules for numeric attributes, such as (Balance c 1) + (CardLoan = yes), which implies that bank customers whose balances fall in a range 1 are likely to use card loan with a probability greater than p. The above rule is interesting only if the range 1 has some special feature with respect to the interrelation between Balance and CardLoan. It is required that the number of customers whose balances are contained in 1 (called the support of 1) is sufficient, and also that the probability p (called the confidence raho) should be much higher than the average probability of the condition being met. Our goal is to realize a system that finds such appropriate ranges automatically. We mainly focus on computing two
Data Mining Using Two%2DDimensional Optimized Accociation Rules Scheme

SONAR System for Optimized Numeric AssociationRules.
We propose an extension of an entropy-based heuristic of Quinlan [Q93] for constructing a decision tree from a large database with many numeric attributes. Quinlan pointed out that his original method (as well as other existing methods) may be inefficient if any numeric attributes are strongly correlated. Our approach offers one solution to this problem. For each pair of numeric attributes with strong correlation, we compute a two-dimensional association rule with respect to these attributes and the objective attribute of the decision tree. In particular, we consider a family R of grid-regions in the plane associated with the pair of attributes. For R 2 R, the data can be split into two classes: data inside R and data outside R. We compute the region R opt 2 R that minimizes the entropy of the splitting, and add the splitting associated with R opt (for each pair of strongly correlated attributes) to the set of candidate tests in Quinlan&#039;s entropy-based heuristic.  We give efficient alg...
An Extension of Van Gelder%27s Alternating Fixpoint to Magic Programs.

Constructing Efficient Decision Trees by Using Optimized Numeric Association Rules.
We propose an extension of an entropy-based heuristic of Quinlan [Q93] for constructing a decision tree from a large database with many numeric attributes. Quinlan pointed out that his original method (as well as other existing methods) may be inefficient if any numeric attributes are strongly correlated. Our approach offers one solution to this problem. For each pair of numeric attributes with strong correlation, we compute a two-dimensional association rule with respect to these attributes and the objective attribute of the decision tree. In particular, we consider a family R of grid-regions in the plane associated with the pair of attributes. For R 2 R, the data can be split into two classes: data inside R and data outside R. We compute the region R opt 2 R that minimizes the entropy of the splitting, and add the splitting associated with R opt (for each pair of strongly correlated attributes) to the set of candidate tests in Quinlan&#039;s entropy-based heuristic.  We give efficient alg...
An Extension of Van Gelder%27s Alternating Fixpoint to Magic Programs.

The Glue%2DNail Deductive Database System Design%2C Implementation%2C and Evaluation
We describe the design and implementation of the Glue-Nail deductive database system. Nail is a purely declarative query language; Glue is a procedural language used for non-query activities. The two languages combined are sufficient to write a complete application. Nail and Glue code are both compiled into the target language IGlue. The Nail compiler uses variants of the magic sets algorithm and supports well-founded models. The Glue compiler&#039;s static optimizer uses peephole techniques and data flow analysis to improve code. The IGlue interpreter features a run-time adaptive optimizer that reoptimizes queries and automatically selects indexes. We also describe the Glue-Nail benchmark suite, a set of applications developed to evaluate the Glue-Nail language and to measure the performance of the system.  1 Introduction  A current focus of database systems research is the design of programming languages and systems to support non-traditional database applications such as computer aided ...
An Alternating Fixpoint Tailored to Magic Programs.
SLD resolution with negation as finite failure (SLDNF) reflects the procedural interpretation of predicate calculus as a programming language and forms the computational basis for Prolog systems. Despite its advantages for stack-based memory management, SLDNF is often not appropriate for query evaluation for three reasons: a) it may not terminate due to infinite positive recursion; b) it may not terminate due to infinite recursion through negation; c) it may repeatedly evaluate the same literal in a rule body, leading to unacceptable performance. We address three problems fir a goal-oriented query evaluation of general logic programs by presenting tabled evaluation with delaying (SLG resolution).
Design and Implementation of the Glue%2DNail Database System.
We describe the design and implementation of the Glue-Nail database system. The Nail language is a purely declarative query language; Glue is a procedural language used for non-query activities. The two languages combined are sufficient to write a complete application. Nail and Glue code both compile into the target language IGlue. The Nail compiler uses variants of the magic sets algorithm, and supports well-founded models. Static optimization is performed by the Glue compiler using techniques that include peephole methods and data flow analysis. The IGlue code is executed by the IGlue interpreter, which features a run-time adaptive optimizer. The three optimizers each deal with separate optimization domains, and experiments indicate that an effective synergism is achieved. The Glue-Nail system is largely complete and has been tested using a suite of representative applications. 
Avoiding Cartesian Products in Programs for Multiple Joins.
. Computing the natural join of a set of relations is an important operation in relational database systems. The ordering of joins determines to a large extent the computation time of the join. Since the number of possible orderings could be very large, query optimizers first reduce the search space by using various heuristics and then try to select an optimal ordering of joins. Avoiding Cartesian products is a common heuristic for reducing the search space, but it cannot guarantee optimal ordering in its search space, because the cheapest Cartesian-product-free (CPF, for short) ordering could be significantly worse than an optimal non-CPF ordering by a factor of an arbitrarily large number. In this paper, we use programs consisting of joins, semijoins, and projections for computing the join of some relations, and we introduce a novel algorithm that derives programs from CPF orderings of joins. We show that there exists a CPF ordering from which our algorithm derives a program whose co...
An anternating fixpoint tailored to magic programs.

Lattice Programming Methodology.
In the framework of symmetric Cournot oligopoly, this paper provides two minimal sets of assumptions on the demand and cost functions that imply respectively that, as the number of firms increases, the minimal and maximal equilibria lead to (i) decreasing industry price and increasing or decreasing per-firm output; and (ii) increasing industry price (and decreasing per-firm output.) In both cases, per-firm profits are decreasing. The analysis relies crucially on lattice-theoretic methods and yields general, unambiguous and easily interpretable conclusions of a global nature. As a byproduct of independent interest, new insight into existence of Cournot equilibrium is developed.
Symbolical Construction of Truth Value Domain for Logic Program.

Prolog Computation Model BPM and its Debugger PROEDIT2.

