International Journal of Software Engineering and Knowledge Engineering

The Ubiquitous Bio-Information Computing (UBIC²) project aims to disseminate protocols and software packages to facilitate the development of heterogeneous bio-information computing units that are interoperable and may run distributedly. UBIC² specifies biological data in XML formats and queries data using XQuery. The UBIC² programming library provides interfaces for integrating, retrieving, and manipulating heterogeneous biological data. Interoperability is achieved via Simple Object Access Protocol (SOAP) based web services. The documents and software packages of UBIC² are available at .

One of the challenges for bioinformaticians is to approximate, in silico, tried and tested research methods used in vitro. One of the problems standing in their way is the lack of a concrete framework for designing and expressing in silico experiments that aim at being isomorphic to in vitro experiments. This paper introduces such a framework in the form of a specification language called ISXL. ISXL projects to biologists a model of in silico experiments that approximates the research method they are most familiar with, as follows. An ISXL-specified experiment (1) conforms to a conceptual model that explicitly captures the basic constituents of experiments in the empirical sciences; (2) may be defined in relation to explicit hypothesis formulation and validation rather than simply taking the form of an evidence gathering process as in alternative approaches; (3) may be long-lived and evolve over time, in the sense that there is built-in support for denoting past versions of specifications, past results, past hypotheses, past validation criteria; (4) may denote other experiments and their constituent parts, thereby reflecting the interrelatedness of scientific processes. Features (1)-(4) above are made possible by endowing ISXL with certain characteristics of a persistent workflow environment. This allows ISXL experiments to be rich in metadata without imposing too great a burden on the biologist. The metadata in turn open the way for ISXL experiments to be capable of introspection and reflection. This paper focuses on describing of ISXL conceptually and syntactically, and indicates how ISXL experiments are given a formal semantics.

Numerous genomic annotations are currently stored in different Web-accessible databanks that scientists need to mine with user-defined queries and in a batch mode to orderly integrate the diverse extracted data in suitable user-customizable working environments. Unfortunately, to date, most accessible databanks can be interrogated only for a single gene or protein at a time and generally the data retrieved are available in HTML page format only. We developed GeneWebEx to effectively mine data of interest in different HTML pages of Web-interfaced databanks, and organize extracted data for further analyses. GeneWebEx utilizes user-defined templates to identify data to extract, and aggregates and structures them in a database designed to allocate the various extractions from distinct biomolecular databanks. Moreover, a template-based module enables automatic updating of extracted data. Validations performed on GeneWebEx allowed us to efficiently gather relevant annotations from various sources, and comprehensively query them to highlight significant biological characteristics.

Indexing protein tertiary structures has been shown to provide a scalable solution for structure-to-structure comparisons in large protein structure retrieval systems. To conduct similarity searches against 53,356 polypeptide chains in a database with real-time responses, two critical issues must be addressed, information extraction and suitable indexing. In this paper, we apply computer vision techniques to extract the predominant information encoded in each 2D distance matrix, generated from 3D coordinates of protein chains. Distance matrices are capable of representing specific protein structural topologies, and similar proteins will generate similar matrices. Once meaningful features are extracted from distance images, an advanced indexing structure, Entropy Balanced Statistical (EBS) k-d tree, can be utilized to index the multidimensional data. With a limited amount of training data from domain experts, namely structural classification of a subset of available protein chains, we apply various techniques in the pattern recognition field to determine clusters of proteins in the multi-dimensional feature space. Our system is able to recall search results in a ranked order from the protein database in seconds, exhibiting a reasonably high degree of precision.

In general biologists are not accustomed to formulating biological problems in the precise mathematical terms that are required to solve the problems analytically or numerically. Although many computational tools for systems biology have been developed recently, our observations indicate that many of these tools are powerful only in the hands of those who know a lot about how to use them. For most biologists, the tools have a protracted learning curve and unfriendly user interface that often diminish their likelihood of being used.

Our long-term goal is to build a knowledge system that allows biologists to synthesize complex biological systems via natural language interactions, and the system is able to generate the corresponding mathematical descriptions so that the often cumbersome communication process between biologists and mathematicians/engineers in formulating complex biological problems in mathematic terms can be performed more easily.

To focus, the first goal in this research is to build a knowledge system prototype that focuses on transport related biological problems that occur from the cellular to tissue level. We address specifically two inter-related problems: (1) Provision of an intelligent system that is capable of automatically synthesizing smaller components into more complex systems; Provision of a user-friendly and natural language interface.

In the era of post-genomics, almost all the genes have been sequenced and enormous amount of data have been generated. Hence, to mine useful information from these data is a very important topic. In this paper, we present a software architecture for finding motifs using genetic algorithm (GA). The new approach can find potential motifs in the regions located from the -2000 bp upstream to +1000 bp downstream of transcription start site (TSS). The mutation in the genetic algorithm is performed using position weight matrices to reserve the completely conserved positions. The crossover in the GA is implemented with specially-designed gap penalties to produce an optimal child pattern. We also present a rearrangement method based on position weight matrices to avoid the presence of a very stable local minimum that may be difficult for operators to generate the optimal pattern. The predicted results obtained from our approach are more accurate than that of Gibbs sampler and we spend less computation time than MEME.