The system for designing the gas separation membrane module, which consists of three engines, was developed. The engine of creating a separation modular structure generates the arrangement coordinates data of a separation unit by GUI, and generates the input file for hydrogen recovery rate calculation. Since this engine can respond to a cylinder and square pillar type module outward form and can respond to unit arrangement of a triangular lattice or a square lattice, the output of the unit arrangement coordinates of general multi-pipe set structures (separation, extraction, heat exchanger, etc.) is possible for it. The engine of performing hydrogen recovery rate calculation uses a commercial computer fluid dynamics (CFD) package, receives input data from a modular structure generation portion, and outputs a convergence calculation result to a file. The engine, which optimizes modular structure, evaluates output results by CFD, and repeats processing in which the structure parameters generated by the genetic algorithm and output results are stored in a database. In the structure design of a separation membrane module, it became clear that the genetic algorithm could optimize efficiently about modular size and element size, or its arrangement.

View full abstract

Chemical Vapor Deposition (CVD) is one of the most important ultra fine manufacturing processes for semiconductor devices. Although simulators are very helpful for developing CVD processes, large calculation costs of the simulators obstruct extensive utilization of them. We showed that modeling of correlations between calculating conditions and calculated results by the simulators radically decreases the calculation costs of them. Therefore, we developed software agent for automatic modeling of calculation processes of the simulators. The agent makes input data for the simulators, and takes output data by operating the simulators. It also makes training data from both the input and output data, and calculates the models of the calculation processes of the simulators by operating general-purpose modeling software. We applied the agent to the simulators for CVD deposition profiles on the substrates with micrometer-sized trenches. The agent successfully made the good training data, and proposed good models of the calculation processes by use of neural networks.

View full abstract

In the field of drug or molecular design, computer is actively utilized for saving development cost and time. Although a large number of computer programs for analysis of chemoinformatics or bioinformatics are available, development of a more intuitive and more multifunctional computer system is desired. In our laboratory, various novel methods have been studied and relative computer programs have been developed. In this study, we integrated these analysis methods into one computer program to develop the molecular design total system 'ToMoCo'. The ToMoCo provides several analysis functions for molecular design or quantitative structure activity relationships such as CoMFA (Comparative Molecular Field Analysis), alignment of molecular structures based on chemical properties by Hopfield Neural Network, automatic structure generation by LigConstructor and region selection by GARGS method in CoMFA and so on. The ToMoCo is developed in C++ language and run under windows 2000/XP or higher. User can perform analysis by above-mentioned method through GUI and can illustrate a result graphically to easily interpret.

View full abstract

We have been developing the chemometrics software package "Chemish". The fundamental calculation methods of chemometrics such as MLR, PLS, QPLS, and PCA have been incorporated in this package. Since Chemish is the chemometrics total system with well-organized GUI, researchers in the field of chemometrics can use it easily and the results obtained by Chemish can be easily interpreted with aid of the graphical GUI. In this study, in order to advance a convenience of Chemish, we have developed the following functions; the function for handling chemical structures, the data modeling function based on the atomic group-contribution method, the function for supporting creation of fragment library, and the function of reverse-analysis to propose new candidate structures. As an example of the above-mentioned functions, the data modeling analysis based on the atomic group-contribution method related to Log P values was performed. As a result of the PLS modeling, the significant model with six latent components was obtained (Q^2=0.80). Then reverse-analysis was carried out to propose new candidate structures that have the desired Log P values, a total of 59 candidate structures were proposed.

View full abstract

Linear fragments of molecular structures have been employed as descriptors in the structure activity relationship studies by the cascade model. A variety of new fragment expressions are developed in order to give better descriptions of characteristic substructures. They include hydrogen-bonded fragments, the omission of element symbols in the middle of a fragment, the lowercase letter notation for atoms in an aromatic ring, and the unified atom treatment for the carbonyl (CO) group. We have to select descriptors out of numerous fragments extracted. Correlation coefficients among categorical attributes are defined. We omit a descriptor when it has a high correlation with other descriptors employed. The resulting descriptor set is efficient and effective in the mining of structure activity relationships.

View full abstract

Coals have been differently formed influenced by the original plants as materials and the heated history under high pressures in the stratum. Therefore, characterization of coals by using the near infrared or infrared spectra is an object of interest for computer aided chemistry. For, example, what would be the independent components of coals spectra ? What would the loading of the spectra mean ? We took about 70 standard coals samples from CCUJ (Center for Coal Utilization, Japan) and measured all the spectra of about 70 standard coals. Further more, the ultimate analysis, proximate analysis and so on are obtained from CCUJ. We analyzed the correlation between the obtained near infrared or infrared spectra and their many parameters by using classification and regression trees, principal component analysis or the other analysis methods.

View full abstract

The Internet has attracted attention as a media for interactive education system for distance- and self-learning systems. Here we report an example of the Flash animations which are available in teaching and learning introductory inorganic chemistry in the university. Flash animation has advantages of being (1) small file size, (2) easy to create, and (3) independent of the platform. In learning inorganic chemistry, it is very important to understand (1) the molecular and crystal structures, and (2) electronic state of non-classical molecules and transition metal complexes, for which graphics and animation are quite effective. Animations featuring (1) diborane B2H6, (2) ion crystal structure, and (3) ligand field stabilization energy are presented.

View full abstract

We have started a project to construct a drug safety information community as a social system, allowing effective circulation of drug safety information among health care consumers, health care professionals, and drug discovery researchers through a shared foundation of knowledge on the internet. In this study, a questionnaire survey on "Drugs and Pregnancy" was undertaken for pharmacists, pharmacy students and researchers to investigate their perception. The response rate was 108/211(51%). Of these respondents, 47.2% were students, 36.1% were pharmacists, and 8.3% were researchers. A number of respondents (95%) were interested in the web site with the information about drug safety during pregnancy. For registration of drugs used during pregnancy and pregnancy outcome, their cooperative trend was found. On the other hand, about 80% of the respondents worried about leakage of personal information. Eighty % of the students and researchers answered that they wanted to use the teratogenicity information for drug design. All researchers answered that it was really beneficial to share the information about reproductive and developmental toxicity, but it must be difficult for pharmaceutical companies. Social, academic and educational activities of information pharmacy, a center of our information community, will gain broad-based acceptance and support of many people.

View full abstract

Conceptual network in the field of Molecular Biology of Cell was usedto analyze how the conceptual network evolves. Conceptual network isgenerated by extraction of terms and by linking terms according tosemantic relationships among terms. Such network represents knowledgeof Molecular Biology of Cell, where a node represents a concept and alink indicates two concepts are semantically related. Three stageswere selected, where the number of concepts are 9,727, 10,983, and 14,373. Analysis based on the number of links per node suggests that the conceptual network evolves nonrandomly. Furthermore, analysis suggest that a new concept is introduced to the conceptual network with single relationship, and progressively conncects to a group of concepts with 15-20 relationships. This number is the threshold of clustering hierarchy. Concepts jumping this threshold seem to become important and/or general concepts.

View full abstract