Applications of data mining to chemistry and medicine are in a starting phase. A difficulty of the application in these fields is from the complex structures of objective data such as molecular structures, genome sequence and therapy history. Another difficulty is from the shortage of human resource having background knowledge in each domain and data maining techniques. Accordingly, the collaboration among domain experts and deta mining engineers is required. In this report, first, present state of data mining applications to industry is outlined. Second, an aplication of a representative data mining method to medicine is shown. Third, more recent and anvanced techniques of data mining is introduced, and their application to Structure-Activity-Relationship analysis is shown. Though these applications, the collaboration among the experts and the engineers is exemplified. Finally, the future trend of data mining is discussed.

View full abstract

[in Japanese]

View full abstract

Some tools of informational chemistry, such as databases, have recently become useful even in the field of experimental chemistry. However, in practical meaning, most of researches of informational chemistry are not essential approaches for solving chemical problems yet. In order to advance practical use in chemistry, establishment of both foundation and applications are necessary. It is a general concept in broad research fields. According to the concept, we are studying several themes concerning with informational chemistry for practical aims in chemistry. This paper describes overview of the researches with their purposes and concepts. It includes quality control study of databases, algorithms for fast computing, graphic interfaces, construction of software library, and developing some chemical representation methods and computer systems for prediction of NMR and chemical reactions for practical use.

View full abstract

Recent advances in molecular computations and simulations of biological molecules, such as proteins, polynucleotides, and lipids, are reviewed. The methods of the classical molecular simulation have been largely refined, and nowadays the classical simulations are extensively performed to analyze the biomolecular systems. The quantum computations, on the contrary, are still in an underdevelped stage. Nevertheless, their recent methodological advances are noteworthy. In particular, the Fragment Molecular Orbital (FMO) method is expected to extend the quantum computations and dynamics simulations of biological molecules owing to its high precision, speed, and parallel efficiency.

View full abstract

Usability of polymer database PoLyInfo has been strongly enhanced by MatNavi and PoLyIndex. The data model is very important for polymer data. We are rewriting the old "Results of object-oriented analysis(class diagram)" for polymer database PoLyInfo. It doesn't match real polymer structures and polymer data structures. For example, As no polymer classes in the diagram have end-group informations, we can't use these polymers as macromonomers or telechelic polymers for other polymer preparations. How can we do the materials designing or molecular designing? The whole structures of the polymers should be shown for materials designs, and now can be represented by the IUPAC Nomenclatures and by the text structures(TXS) developed for PoLyInfo's data treating.

View full abstract

We have been developing a CAST/CNMR system for highly accurate prediction of NMR chemical shift values. In the CAST/CNMR system, a modified CANOST method is used for a planar structure coding. The calculation time for generating of the modified CANOST codes is sufficiently reduced with application of algorithms in graph theory, e.g., breadth fast search. This approach provides the fast generation of canonical coding. The implementation of the fast algorithms advances the practical use of the CAST/CNMR system.

View full abstract

Correlation of the octane number (ON) of heptane and octane isomers with various topological indices, Z, w, p, and B, was studied. One of the best results of regression analysis is: ON=76.389+8.179(0.3B+1.1p-0.6Z) with r=0.982 for octane isomers. From these analyses one can concluded that in order for a gasoline isomer is to have high ON, or to be burnt without knocking, it should be highly branched (small Z or low boiling point), its statistically averaged shape should be spherical (large B), and it should have a compact structure (large p or high liquid density). From these results several candidates of nonane isomers whose ON is expected to be higher than 100 are suggested.

View full abstract

We developed a system for classifying G-protein coupled receptors (GPCRs) to families, and obtained results with high precision for both aligned and non-aligned amino acid sequences. The system developed in this study employs self-organizing map (SOM), which is a kind of artificial neural networks and its output neurons learn to respond individually to distinct input patterns so that nearby neurons respond to similar inputs. That is, SOM performs clustering feature for a given data set. As a result, sequences of amino acids are partitioned to functional family areas (such as amine, peptide, olfactory, etc.) on the map. Since the system predicts families of new sequences and orphans with high precision, it is expected that it works as a strong tool for classifying sequences of amino acids. The learning process were carried out for the various combinations and normalization degrees by using 50 50 square grid neuron space, and the basic batch method. The precisions of the classification are 97~99% for the training set, and 96~98% for the test set. The results suggest that the method works very well for non-aligned data, suggesting that SOM is a promising method for non-aligned data.

View full abstract

SYNSUP is an organic synthesis design system, which has been developed by our team since 1984. For a given target molecule it carries out a depth-first search in the synthesis tree non-interactively. To make it wide spread among organic chemists as a handy tool for synthesis planning, a user-friendly environment has been established. Once the user prepares the input data and submits the job by e-mail using CMBedit, our original graphic user interface, they have only to wait to receive the reply mail. A file attached to the mail includes proposed routes with corresponding literature references and purchasable chemicals' information. Proposed routes are examined using the output interface. In the map view mode, one can see the outline of the tree of synthesis routes. A click on the reaction arrow causes brief comments and the corresponding references to be shown in a separate window. When a reactant is purchasable the structure is displayed as surrounded by a box. One click produces the supplier's name and the catalog price. Communication between the system and the users is managed by a unique procedure, a combination of OpenPBS (Portable Batch System) and several Perl scripts.

View full abstract

We have been studying how to use the transition states data base (TSDB) for developing new synthesis routes. One of the usages of TSDB is to confirm whether or not a new route designed by a synthesis route design system (SRDS) such as AIPHOS/KOSP or TOSP is possible to synthesize a target compound. For this purpose, we have to take steps as follows; (1) Design the routes using SRDS; (2) Find similar reactions from TSDB by using FIND_TSINFO; (3) Make the starting geometry which is derived from that of TSDB to calculate transition states by using TS_Search; (4) Calculate transition states by using Auto_PTOD. While there is little possibility that the TS information for a new route exists in TSDB, it is likely to find data of reactions similar to that of the new route created by SRDS. The information of the similar reactions are very helpful to find the TS for the synthesis routes from SRDS. In the present study, we investigated how to use TSDB for synthesis routes of 1-(cyclohex-3-enyl)ethanone created by AIPHOS/KOSP and TOSP.

View full abstract

The concept of RS-stereoisomeric groups is proposed to integrate stereochemistry and stereochemical nomenclature, where the three pairs of concepts, i.e., chirality/achirality, RS-stereogenicity/RS-astereogenicity, and sclerality/aclerality, are ascribed to the three normal subgroups of an RS-stereoisomeric group.

View full abstract

The actual measurements of safety data for existing chemicals have not proceed for many years in Japan as well as in OECD countries, because it takes huge expenses, time and labor.In order to solve those problems, the authorities for the management of chemical substances in each country have been making their efforts to utilize the prediction by (Quantitative) Structure Activity Relationships [(Q)SAR] in place of actual measurement. As for the biodegradation data, Japan have been accumulating reliable data for many years, which is utilized globally for the research and development of (Q)SAR systems.NITE focused its attention to separate biotic reaction and abiotic reaction occurring in the biodegradation measurement system. Those reactions are based on quite different mechanism, it would be useful to separate abiotic reaction from biotic reaction to improve the predictivity of biotic reaction.NITE started the program of developing the prediction system for abiotic hydrolysis of chemicals based on the molecular orbital method.

View full abstract

The external validations of four biodegradability predictive models (BIOWIN5, BIOWIN6, CERI model and CATABOL) were conducted using test data of 1121 new chemicals under the Japanese chemical substances control law (CSCL). The rates of the number of chemicals that are correctly classified into ready and not-ready biodegradable are 76 % (BIOWIN5), 77 % (BIOWIN5), 81 % (CERI model) and 81 % (CATABOL), respectivery. Following common tendencies in prediction were found in all models: 1) The predictivities of ready biodegradability (56-75 %) are lower than the predictivities for not-ready biodegradability (82-85 %). 2) The predictivities for aliphatic chain compound are reratively low (66-72%). 3) The predictivities for compound with moiety that react to water such as ester are reratively low (66-72%).

View full abstract

Recently, it is required to investigate decomposition of toxic substances emitted to nature in order to avoid environment disruptions. However, we cannot measure rates of reactions for all the compounds since there are enormous numbers of toxic compounds. In order to predict whether or not a toxic compound easily decomposes to others at the environmental conditions, theoretical calculations can play an important rate in calculating information such as activation energies. It is because values are closely related to rates of decomposition of compounds. In the present study, hydrolysis of esters in the acidic conditions was investigated at the B3LYP/6-31G* level of theory. There are many esters which are toxic and widely used in industrial field. Theoretical calculations did not locate any tetrahedral intermediates which many text books adopted as key intermediates. We offered an alternative mechanism which was an activation barrier of 39.2 kcalmol-1. This result is consistent with experiments.

View full abstract

The number of nodal regions can be used as an index for characterizing molecular orbitals. A computer program has been developed to count the number of nodal regions, based on the labeling and contraction algorithms. This program is applied to the water molecule, the hydrogen atomic orbitals, and the Rydberg excited states of ethylene. Since the number of nodal regions is independent of the coordinate system, the method is applicable even when the molecular structure changes drastically as in bond rotation or bond elongation. Changes of nodal regions with bond elongation are investigated for carbon monoxide. A prescription for problems arising with basis set expansion techniques is also given.

View full abstract

Using ab initio quantum chemical methods, we have investigated Ruthenium polypyridyl complexes, which are key components of dye-sensitized solar cells with the TiO2 semiconductor photoelectrode and the redox electrolyte solution. The optically allowed singlet excited states of Red-dyes are calculated by SAC (Symmetry adapted cluster)-CI (Configuration interaction) method with PCM (Polarizable continuous model) method to include the solvation effect. We have investigated the effects of the solvation and pH change effects to the absorption spectra. The ab initio calculations quantitatively well reproduced the experimental facts for the solvation and pH change effects. With the assistance of theoretical and computational chemistry approach, the reproduction of the experimental data with the ab initio calculations leads us to design of high-performance dye-sensitized solar cells without a lot of experimental trial and shorten the development periods for the optimal materials and the optimal conditions.

View full abstract

To elucidate delicate properties of proteins with higher accuracy, we have developed a density functional (DF) program, ProteinDF which can treat the whole protein as a molecule. Now, we are developing as a ProteinDF system extending the functions and performances under the "Frontier Simulation Software for Industrial Science" project supported by the IT program of MEXT. When the first all-electron calculation of protein was achieved in 2000, the convergence of DF calculation was extremely difficult and the computational time was extraordinarily long. In this study, it was shown that all-electron DF calculation on proteins could be attained by a practical time without any trial-and-error. We will argue about the whole system image and its future view with some examples of developments and calculations.

View full abstract

We propose Atomic Schwinger-Dyson formalism(ASD method) and renormalization under an external field in this paper, which is the nonperturbative and finite relativistic quantum field theory for a finite baryon and electron density. ASD can be applied to many of electron systems and electron matters. ASD consists of coupled Dyson equations of electrons and photons. It is shown that an electromagnetic field of ASD is separated into two parts, which are a mean field (classical field) and a fluctuation (quantum field). Because it includes self-energies in nonerturbative way, higher-order correlations beyond Hartee-Fock approximation are taken into account. In order to prepare for numerical calculations, we give a new explicit expressions of photon self-energies and electron self-energy, which are introduced for photons and electrons based on the particle-hole-antiparticle representation (PHA) of Atomic Schwinger-Dyson formalism (ASD). The PHA representation describes exactly the physical processes such as particle-hole excitations (electron-hole) and particle- antiparticle excitations (electron-positron).

View full abstract

It is critical problem in nano-technology and bio-science to improve the quality of classical molecular simulations. One of the key technologies for the practical solution must be the accurate evaluation of intermolecular interactions to be able to represent the crystal structures. On the other hand, the recent approach to multi-scale simulation hybridized quantum MO and classical MM techniques is expected in actualization of mesoscopic and bio-system simulation. However, it seems that combination of two different methodologies causes discontinuity of the theoretical boundary line if it is performed without any refinement of molecular force field. In this conference, we will introduce our recent works in developments of crystal force field and incorporation of QEq-GB calculation into the force field. Requirements of the next generation of molecular force field will be also discussed.

View full abstract

0rds153701el**p1033Detailed investigation on the origin of the acidity of the a-protons of a set of the carbonyl molecules were carried out on the basis of properties of the localized molecular orbital. An anomalously high acidity of Meldrum's acid, as compared with those of dimedone and dimethyl malonate, is one of the well-known but unresolved issues. The well-localized s orbitals of the C-H bonds at the a-position of the carbonyl groups can be obtained with the reactive hybrid orbital (RHO) theory. We found that the energysid10911196 levels of the unoccupiedRHOs of the C-H moiety of Meldrum's acid and other carbonyl compounds showed a good linear correlation with the experimental deprotonation energies. This is probably because the deprotonation reaction to form the proposed naked anions in a polar solvent is a highly endothermic process, in which the thermodynamic energy differences between the neutral molecules and the corresponding anions approximately coincide with the activation energies.

View full abstract

Chemical shift data of platinum-195 for about 7000 species for coordination complexes, organoplatinum and cluster species were compiled to construct the computer-readable database for coordination chemists and other usage. Several spread-sheet type software (Lotus-123, Excel-97 etc) were used to compare and refine the original data, because these contain occasionally several peculiar descriptions, which seems to be impossible experimental procedures. Chemical shift data were arranged with two customary used standards, namely [PtCl6]2- and [xi] (resonant frequency at 2.38 Tesla). .Bibliographic data (First author and Journal data) were added in a compressed form.

View full abstract

The 13C and 1H NMR spectral database for natural products, CH-NMR-NP, started to build from May 2002 has been developed to increase the number of the compounds. The chemical information, NMR data and the spectral assignments are compiled from the literatures. The number compiled in CH-NMR-NP is 4,100 at the end of September, 2004. The addition of the data is continuing. The CH-NMR-NP system for search compounds and display the data has been also developed on the web system by using a DBMS "mysql" and the Java applet. Recently, the substructure search was introduced in addition to the search by text format such as the name of compounds, molecular formula, 1H and 13C shifts and others. The function of the substructure search was achieved by using the chemical-structure drawing software "MarvinSketch / MarvinView" and search engine "JChem" developed by ChemAxon. The matching of the database "CH-NMR-NP" and the software was good.

View full abstract

Integrated Spectral Database System for Organic Compounds (SDBS) is a spectral database that contains six kinds of spectra of Mass, 1H NMR, 13C NMR, IR, Raman and ESR under a compound dictionary. Since April 1997, we open the system to public through the Internet freely; URL is "http://www.aist.go.jp/RIODB/SDBS/". Since introducing SDBS to the Internet, the number of accesses has been rising rapidly; total number of access at the end of July, 2004 was almost 85 million. The enormous accesses and the remarks from all over the world prove the importance and necessity of SDBS. We started to analyze an access log of SDBS for obtaining ideas of improving SDBS and to learn users and usage of SDBS. 1H NMR is the most popular spectrum to be accessed among our six kinds of spectra. Users access compounds having between six and ten carbon atoms, the access to compounds having more than 40 carbon atoms are also popular compare to the number of compounds compiled in SDBS.

View full abstract

We have been developing three different types of organic synthesis design systems (AIPHOS, KOSP, TOSP). On the other hand, in order to evaluate the synthetic pathways proposed by these systems, the reaction evaluation system based on prediction of reaction by-product also has been developed. This system can evaluate synthetic pathway with attention to the reaction condition and the safety of by-product. Then, we developed the Client/Server (C/S) system integrated with three organic synthesis design systems and the reaction evaluation system. This C/S system can operate three organic synthesis design systems and the reaction evaluation system through the common user-interface. In this work, we developed the function of reaction prediction between two reactants and the function of sorting reaction pathways with attention to reaction site or reaction condition. Consequently, the reaction evaluation system becomes possible to evaluate the reaction pathways generated by organic synthesis design sysytems with attention to reaction by-products or reaction condition.

View full abstract

In the exploratory synthetic research in the medicinal chemistry, it aimed at considering as an aid of a synthetic design by predicting synthesizable new compounds comprehensively using the knowledge bases of organic reactions. For this purpose, it used as forward-reaction prediction engine the Reaction Estimation System Using Knowledge base, RESUK, based on the synthetic reaction knowledge base, currently developed at Funatsu laboratory. In this research, we have developed the system which predict the comprehensive synthesizable new products by multi-step reactions from arbitrary start compounds, and visualize the predicted synthetic tree. As a problem assumed here, the number of the predicted products and their reaction schemes becomes huge (explosion of combination). For this reason, the post-evaluation system which evaluates the predicted products and reaction schemes by indices specified by the user, such as the structure feature of a product and yield, performs prioritization and sort. Furthermore, we developed the visualization tool for displaying intelligibly a vast quantity of products and their synthetic tree.

View full abstract

The computer system has been developed to convert chemical name of organic compound to the 2D-structural formula. This report describes full support of nomenclature of natural products and related compounds recommended by IUPAC 1999 on this system. Stereochemical information on a chiral center atom of stereoparent, fundamental parent structure of natural product, is expressed by means of rotation and bond data obtained as follows. The atoms adjacent to a chiral center are arranged in order of increasing locant number. Rotation of the first three atoms in this order around the center is evaluated to be clockwise or anticlockwise on the plain of the standard structural diagram. Bond directed from the center to the 4th atom is evaluated to be up or down. Rotation around an achiral ring atom is also evaluated similarly. Ring atoms of stereoparent are classified into bridge atoms and base ring atoms. The bridge atom is here defined to have a stereochemical bond from the chiral center. Stereochemical indication alpha or beta must be given for the bond from chiral center to bridge or chain atom shown on the diagram of stereoparent structure. Notation about stereochemical change caused by alpha or beta is discussed. Modifications by nor, homo, seco, cyclo and abeo as well as ent and rac are also discussed. The final process to represent up and down bondage on the 2-D structural diagram is discussed in detail.

View full abstract

The enumeration of organic reactions and the evalution of their possibilities are important to comprehend the scope of organic reactions. This problem can be treated by a computer of high speed and of large free memory. By representing organic reactions as Imaginary Transition Structures (ITS), we can extract structural information and reaction type information in an effective way. In this study, we use an algorithmic combinatorial analysis through a computer and enumerate unique reaction graphs which form 5- and 7- membered ITS rings. Resultant unique reaction graphs are stored as EPS files. As a result, we can enumerate reaction graphs with considering valences, although this type of enumeration is impossible in the mathematical combinatorial analysis.

View full abstract

XyM Notation is a kind of linear notation systems for representing chemical structures in the XyMTeX system, which is a package of macros proposed by Shinsaku Fujita for typesetting chemical structures using LaTeX system. Fujita et al. also reported the XyMML (XyM Markup Language) that enabled to embed XyM Notations in HTML documents, and the XyMJava System that is a Java applet for drawing chemical structures on a CRT display decoding the XyM Notation in the HTML documents. Those reports showed that the XyM Notation is useful for the communication of chemical structures through the Internet, but XyM Notation is not familiar for ordinary experimental chemists. To relieve this disadvantage, we developed a program for assisting to construct chemical structures by XyM Notation. By using this program, a user can build chemical structures on the CRT display by combining structural parts, and then the program generates XyM Notation.

View full abstract

The goal of our project is to develop an interactive visualization method and to implement a human-centric domain-oriented graphic library for the method to represent molecular information that helps chemists in synthetic design, reaction prediction, and molecular design. The molecular information is derived from the FRAU system, which characterizes molecules based on electrostatic and steric interactions with a probe as a pseudo-reactant. The visualization library uses a 3D multimedia open-source graphic library, JUN as a general basis and extends it by applying interaction design techniques by taking into account chemists' cognitive process. This paper outlines the project with our initial attempts of the visualization.

View full abstract

The study of intermolecular interactions is a useful step toward gaining a basic understanding of the dynamic behaviour of molecules in various environments. Especially, interactions of the ammonia (NH3) molecule with other molecules provide valuable information on protein-protein interactions or protein-solvent interactions in biological systems. From this point of view, high-level ab initio calculations [MP2] using correlation consistent basis sets of Dunning [aug-cc-pVTZ] were carried out to study the structures and stability of He2/NH3, Ne2/NH3 and Ar2/NH3 Van der Waals trimers and to clarify the nature of three-body interactions. For the most stable structure of He2/NH3, each He atom is located midway between two of the NH3 hydrogen atoms. On the other hand, the global minima on the potential energy surface of Ne2/NH3 and Ar2/NH3 occur at geometries in which a N-H axis of NH3 directed toward one of the rare gas atoms and the other rare gas atom midway between two of the NH3 hydrogen atoms. He-He interaction in He2/NH3 is considerably weaker than in He2 dimer. However, for the Ne2/NH3 and Ar2/NH3, the interactions between two rare gas atoms are almost as strong as in rare gas dimers.

View full abstract

Recently, the experimental detection and characterization of Van der Waals trimers is a subject of growing interest. In the present study, the structures and energies for the Ar-(NH3)2 Van der Waals trimer have been studied at the level of second-order M?ller-Plesset perturbation theory (MP2) using correlation consistent basis sets of Dunning [aug-cc-pVTZ]. For the (NH3)2 dimer, intermolecular potential energy surfaces were calculated at the several gemometries. The most stable structure exibits an asymmetrical geometry, in which a value of 3.24 angstrom is obtained for the N-N distance and the two C3 axes of the NH3 molecules and the N-N axis does not exist in the same plane. For the Ar-(NH3)2 trimer, two stable isomers were located and characterized. The most stable isomer exhibits the structure in which an Ar atom interacts with two NH3 molecules. In the other isomer, an Ar atom interacts only one NH3 molecule. The details of the potential energy surfaces are discussed.

View full abstract

The d6 cobalt(III) complex does not display a striking structural deformation from octahedral geometry. One of the exceptions is the cobalt(III) carbonato complex with a cabonato chelate, which shows a great deviation of the O-Co-O bond angle of 70 degree from the octahedral one (90 degree). This is due to the steric restriction of the chelate carbonato ligand of a four membered ring. This deformation should supply some strain for the cobalt(III) complex and influence the structure around the cobalt ion and the reactivity concerning this complex. From this viewpoint, some experimental approaches have already been carried out. One of the examples is the trans influence of a chelated carbonato. Any distinct influence was not detected in the Co-N bond length trans to the Co-O bond but the N-Co-N bond angle trans to the O-Co-O angle becomes larger than the octahedral angle (90 degree). In this work, we determined the structures, bond lengths and angles for several carbona to cobalt(III) complexes by using the DFT calclation at B3LYP/6-31G* level. After comparing the DFT and X-ray structures of these complexes, we discussed about the calculation reliability of this method for the cobalt(III) complexes having a strain in the structure.

View full abstract

It is well known that gold can be dissolved in aqua regia. Recently, reversible dissolution/deposition of gold has been reported in organic solvent with excessive iodine. Effective dissolution occurs when comparable amounts of I3- and I- are present. To examine the geometries and electronic structures of gold halides, which is closely related to the dissolution/deposition reactions, ab initio MP2 and DFT B3LYP calculations were carried out using nonrelativistic, quasi-relativistic, and relativistic effective core potentials (ECPs). Geometries of diatomic molecules are compared experimental values. Relativistic effects shorten the bond lengths, which is in agreement with the previous works, by approximately 0.2 A. There is not much difference in the calculated bond lengths between MP2 and B3LYP. Optimized geometric parameters of linear triatomic anions of gold halides are also compared with the experimental ones. Studies on geometries and electronic structures of pentaatomic anions of gold halides and polyhalides are in progress.

View full abstract

Cyclazines, of which unique tricyclic structures and perfectly conjugated perimeters of sp2 hybridized carbon atoms, have attracted much attentions from theoretical chemists in analyzing the essential nature of aromaticity/anti aromaticity. Among a series of peripheral conjugated compounds, cycl[3.2.2]azine has been taken as an ideal [10]pai perimeter model bearing an isoelectronic system with [10]-anulene, aimed to assess its aromaticity from a topological viewpoint. Its physicochemical properties as well as chemical reactivity, however, have not yet satisfactory elucidated. In particular, the UV/Vis absorption bands have not fully analyzed in a theoretical way. Here we report the novel synthesis of cycl[3.2.2]azine derivatives and their electronic spectra, followed by a series of molecular orbital calculations (TD-DFT and MS-CASPT2) aimed to characterize the intense absorption peaks. Our results have shown that the absorption peak position associated with the lowest pai-pai excitation has been correctly predicted with both TD-DFT and MS-CASPT2. The corresponding oscillator strength, however, has been considerably underevaluated using TD-DFT. This abnormality has been corrected using highly accurate MS-CASPT2.

View full abstract

It is well known that frontier orbitals of molecules are important to understand chemical reactivity. We have published papers on relations between frontier electron densities and toxicities of dioxins and dibenzofurans. In this work, we investigate the relations between frontier orbital similarities and toxicities of dibenzofurans. Highest occupied molecular orbitals (HOMO) of 136 congeners of dibenzofurans are calculated by ab initio Hartree-Fock molecular orbital method. HOMO of 2,3,7,8-Tetrachlorinated dibenzofuran (2,3,7,8-T4CDF) is used as reference orbital to estimate the frontier orbital similarity, which is overlap integral between HOMO of 2,3,7,8-T4CDF and HOMO of the other dibenzofuran. We have found that frontier similarities of dibenzofurans are affected by the orbital symmetry of HOMO of the reference molecule. We also define atomic similarity of frontier orbital. Multivariate statistical analysis, especially, principal component analysis is performed for the data sets of atomic similarities ofdibenzofurans. There is a correlation between principal component scores of the first principal component and toxicities of dibenzofurans.

View full abstract

Irinotecan (CPT-11) is a widely-used potent antitumor drug that inhibits mammalian DNA topoisomerase I (Topo I). However, overexpression of ABCG2 can confer cancer cells resistance to SN-38, that is, the active form of CPT-11. In the present study, to develop a platform for the molecular modeling to circumvent cancer drug resistance associated with ABCG2, we have characterized a total of fourteen new SN-38 analogues by some typical properties, which were evaluated by molecular orbital (MO) calculations and neural network (NN) QSAR technique. The NN was applied to estimate hydrophobic properties (LogP) of the analogues. Thereafter, the electrostatic potential (ESP) and the salvation free energy (dG) were evaluated by MO calculation. These indexes were found to be well correlated with the drug resistance ratio experimentally observed in ABCG2-overexpressing cells. It is suggested that hydrophilic analogues carrying OH- or NH2-groups are good substrates for ABCG2 and therefore exported from cancer cells. In contrast, SN-38 analogues with Cl atom have similar LogP values and high affinities toward the putative active site of ABCG2, however they were not substrates of ABCG2. From these results, it is strongly suggested that hydrogen bonds are critically involved in the transport mechanism of ABCG2.

View full abstract

The reaction rate with hydroxyl radical (OH) is important, since the tropospheric lifetime of a chemical substance is mainly determined by hydrogen abstraction of OH radical. We proposed the new economical method for estimating the reaction rate on the basis of the transition state theory and the molecular orbital method. Our estimation consists of the following three steps. (1) The structure and vibrational frequencies of a reactant and transition states are calculated by MP2/6-31G** method. (2) The energy of a reactant and transition states are corrected by MP-SAC2 method. (3) The reaction rate constants are calculated by the formula of bimolecular reaction. The parameter F2 of MP-SAC2 method was determined to be 0.865 so as to reproduce the observed rate constants of methane, ethane and their fluorides. In order to show the validity of our method, we applied the method to three fluorinated propanes. We calculated the rate constants of hydrogen abstraction and compared them with experimental values. The calculated rate constants agree with observed values within estimation uncertainty of +/-0.4 in log k.

View full abstract

DFT calculations on the proton transfer in the dimers of the enol form of conjugated 1,2-diketones, tropolone and 2-hydroxycyclopentadienone, were performed. Although the two conjugated carbon rings in both dimers were essentially coplanar in the ground state, they were severely twisted to each other in the transition state of the simultaneous, intermolecular double proton transfer process. From close examination of the hypothetical coplanar structure for the transition state the twisted transition-state structures were ascribed to the tendency of the linear arrangement of the intermolecular O...H...O hydrogen bonds. The coplanar structures of the ground state were attributed to that the intramolecular hydrogen bonds are important than the intermolecular hydrogen bonds, which were supported the results that the dimerization energies for the tropolone and the 2-hydroxycyclopentadienone dimers (5.9 and 8.9 kcal/mol, respectively) were much smaller than that (16.4 kcal/mol) for the benzoic acid dimer, which is a structural isomer of the tropolone dimer. The proton transfer process accompanied with inversion of the one carbon ring was also discussed.

View full abstract

In order to represent a lot of crystal structures of organic compounds including biomolecules and functional materials, we have been started developing a new crystal force field. Atomic pairs having strong short-range interactions have been extracted from Cambridge Structure Database (CSD), and evaluated by Modified Lippincott-Schroder (MLS) function. Intermolecular potential force field has been defined by the combination of the MLS function and general functions of nonbonded interactions (vdW and electronic interactions), and the function parameters have been optimized by using a potential parameter optimization program (KGA99). Using this force field optimized lattice constants 150 crystal structures selected from CSD are good agreement those of experimental data.

View full abstract

To study structure and spectroscopy correlation, Chlorophyll (Chl) a/b dimers, in a sandwich configuration, were generated from the X-ray crystallographic data for a water-soluble Chl-protein complex (WSCP) from Lepidium virginicum. The dimers were optimized with the molecular mechanics method using the MM2 parameters and the energy and oscillator strength of their electronic transitions were calculated with ZINDO using the INDO/1 parameterization. The transitions for all Chl dimer were split into high- and low-energy components with respective large and small oscillator strengths. These calculated results were explained in terms of excitonic coupling in a sandwiched dimer and the coupling strength was estimated to be 190cm-1 for Chl a - Chl a, 150cm-1 for Chl b - Chl b, 80cm-1 for Chl a - Chl b. On the basis of the results, the spectral shape of the Qy band for the WSCP was reproduced by Gaussian fitting in the constraint of a Chl a/b molar ratio of 1.8 determined from exp.. The fitting also provided the composition of the dimers : Chl a - Chl a 55% , and Chl b - Chl b 30% and Chl a - Chl b 15%.

View full abstract

The fragment molecular orbital (FMO) method is one of the approximated molecular orbital (MO) method developed by Kitaura et al. In the FMO method, a target molecule is divided into small fragments and MO calculations are performed for each fragment and fragment pair to obtain the total energy and one-electron properties of entire molecule. Therefore the FMO method makes it possible to calculate the electronic structure for large molecules such as proteins, enzymes and DNAs. In the FMO method, the positions to divide bond in the fragmentaion are very important because they affect the accuracy of the obtained electronic structure. The fragmentation for proteins and DNAs can be performed routinely since the bond-dividing positions for them have been determined experimentally. For chemical species except proteins and DNAs, however, the bond-dividing position has not been revealed, yet, which is the main reason to limit the application of the FMO method. In this study, we inspects the accuracy of total energy and electron density by FMO method by way of fragmentation formats of catechine molecule, which is known of physiologically active substance. The accuracy varies with fragment formats and there is strong correlation between total energy and electron density. There exist some fragmentation formats where total energy differences are less than 10kcal/mol and they give correct electron densities. Therefore we think that there are some appropriate formats besides conventional fragmentation format.=

View full abstract

We are developing a density functional method program ProteinDF for all-electron calculation on proteins. ProteinDF adopts the fuzzy cell method to evaluate exchange-correlation potential so that grids and their weights are estimated for numerical integrals before self-consistent field (SCF) calculation. Because this task is depending on the 3rd power of the number of atoms, it becomes one of the time-consuming routines when the system size is very large. Recently we achieved the 26766 canonical orbitals calculation of a protein, where the grid generation routine cost 22% of all calculations. In order to calculate larger proteins and achieve structure updated calculations such as geometry optimization and ab initio molecular dynamics, it is very important to improve this routine. In this study, we modify the loop structure in the grid generating routine and improve the computational cost by compensating memory. As a result, the performance is increased 4 times, and the share of the grid generation was reduced to 5% for 26766 orbitals calculations. Now, we modify the fuzzy cell algorithm for the further speed-up.

View full abstract

The structure-based drug design (SBDD) is widely accepted as a powerful tool in obtaining novel compounds that specifically bind the corresponding target molecule. The accurate molecular structure of the target molecule and a suitable method that can properly evaluate the interaction between the drug candidate and the target molecule is highly required to perform SBDD effectively. A molecular orbital calculation has been regarded as one of the most useful methods for the purpose. The recent advances in computer hardware and software technologies are now making this method a possible practical tool for drug design. Since the optimization of all atoms in the target molecule is particularly desirable for SBDD, appropriate protocols for the calculation should be developed. Using a semi-empirical molecular orbital software named LocalSCF, we have investigated several conditions appropriate for this purpose and developed a suitable protocol.

View full abstract

Hydrogen bond and hydrophobic interaction are important to determine three-dimensional structure of protein. We analyzed conformational change of oligopeptides caused by intra-molecular hydrogen-bond formation and/or deformation using ab initio MO-MD method based on the Hamiltonian Algorithm at the HF/3-21G level. All initial conformations were assumed as alpha-helix structures. A glycine pentamer has two intra-molecular hydrogen bonds an initial alpha-helix conformation. One of these hydrogen bonds was formed between 1- and 4-residues, the other was between 2- and 5-residues. Both hydrogen bonds were deformed at around 3000 steps in an energy trajectory and new three intra-molecular hydrogen bonds were formed simultaneously. In the case of alanine pentamer, two hydrogen bonds between similar residues as a glycine pentamer exist at an initial conformation. These hydrogen bonds were deformed at earlier (about 1500) steps than the glycine pentamer. After 4000 steps, new hydrogen bond between 1- and 5-residues was furthermore formed. We also carried out the analysis for typical acidic oligopeptides such as asparatic and glutamic acids pentamers. Our proposed ab initio MO-MD method is efficient to theoretically analyze dynamical process of hydrogen-bond formation/deformation in oligopeptides.

View full abstract

It is important problem in a protein science to detect active sites on protein structures. Recently some automatic detection methods were presented; finding a similar structure to known protein active sites, finding pockets and/or cavities on protein surfaces, and so on. We already presented a method to detect amino acid residues, which construct a function site, using the similarity in 3-D structure of same function proteins. In this study, we applied our method to protein pairs of serine proteases for clear the scope and limits of the method. Serine proteases are classified into subgroup by amino acid sequence similarity and into group by folding types. The extraction of active sites from same subgroup, different subgroup but same group and different group serine protease pairs were investigated with various threshold values of allowance distance between amino acid residues. By considering the extracted amino acid residues, it is found that the method could be obtained the true active site residues using pair of proteins with different amino acid sequences but same folding type.

View full abstract

We have reported that classical methods such as the principal component analysis are insufficient for profiling illegally distributed drugs. Then we have used Livingstone-type hierarchical neural networks and the neural independent component analysis to classify confiscated methamphetamine. However, the two methods did not always show sufficient results for profiling such methamphetamine. In this study, other two methods, SOM and kernel principal component analysis, were applied in order to classify the confiscated methamphetamine using GC-MS data of impurities. To classify the data using SOM, not only winner neurons but also all the weights of connections were used like fingerprint matching. Thus, better classification results, which are consistent with the result from a synthetic investigation, were obtained. However, since SOM is a classification method, it is also desired to apply nonlinear component analyses. Kernel principal component analysis is one of such nonlinear analyses; it uses higher-dimensional feature space in order to find appropriate principal axes. This method gave the classification result that is easier to understand than SOM.

View full abstract

The use of quantitative structure - property relationship to predict the logarithmic aqueous solubility coefficients, logSw, of 136 congeners of polychlorinated biphenyls, PCBs, known as environmental contaminants is reported. Principal component regression, PCR, and computational neural networks, CNNs, were used to develop models using a set of information-rich descriptors, which were encoded information about molecular topology, geometry, and net charge at atoms calculated by using MOPAC 93 rev. 2 software package. We compared the square of the regression coefficient, r2, values obtained from two models. For the congeners, both the models seem to be appropriate for the prediction, since the r2 values were 0.953(PCR) and 0.982(CNNs), respectively. However, it was found that they were insufficient for predicting logSw of chlorinated biphenyl, CB, isomers. Therefore, we reconstructed prediction models for the CB isomers by using them. It was elucidated that CNNs derived! adequate models for predicting logSw of CB isomers, and that they were markedly preferable to the models derived from PCR judging from the r2 values.

View full abstract

Nematic liquid crystal is of significant industrial importance due to the potential for liquid crystal display (LCD) application. There are number of different display driving modes, and each of them requires different sort of liquid crystalline properties. In order to find out liquid crystals that are the most suitable for the applications, it should be important to know a correlation between the chemical structures and the properties. Among the properties, however, in every case, the most important and common demand is a wide and stable nematic phase. Up to now it has been recognized being extremely difficult to know exact chemical structures showing tendency to exhibit a nematic phase by other calculation methods i. e., molecular orbital calculations because the mesomorphic potential might be derived from the sensitive inter-molecular interaction.Applying Neural Network we have succeeded in development of a methodology estimating a relation between the chemical structures and the mesomorphic potential especially of the nematic phase. With 200 representative liquid crystals having wide variety of chemical structures, we have evaluated a relation of the chemical structures and the mesomorphic properties, and as the result the suitable Neural Network descriptors have been obtained. The developed methodology allows us to know if a chemical structure exhibits a nematic phase in 92% of accuracy without synthetic work.

View full abstract