Nanotechnology is a basic science for materials development that covers almost all artificial materials. Although research in materials science has emphasized differences among materials, similarities should also be studied from the viewpoint of nanotechnology. To make knowledge in one domain easier to use in other domains, the knowledge must be generalized into common categories that can contribute to the realization of a structured knowledge platform. Materials design focuses on understanding the relationship between processes, (material) structures, and functions. Materials science is divided into disciplines based on the properties of bulk materials. However, common rules that govern phenomena at nano-scales are eliminating those boundaries between disciplines. In order to support more effective education, research, development, and manufacturing in materials science, a Japanese national project for structuring knowledge of materials nanotechnology is underway. Examples of the effect of structuring knowledge on the relationships between processes and material structures are shown for an advanced device such as fuel cell. A knowledge platform, into which this generalized nanotechnology knowledge is to be integrated, is outlined in this paper. The key purpose of structuring knowledge is to stimulate idea generation based on a fundamental and general understanding of underlying mechanisms.

View full abstract

The results of various genome projects have shown that up to 30% of human proteins occur in cell membranes. Membrane proteins play crucial roles in many biological functions and are of key importance for medicine. Over 50% of commercially available drugs target membrane proteins. In spite of the abundance and importance of membrane proteins there are only 100 unique membrane protein structures in the Protein Data Bank. To address the technical bottlenecks preventing the structure determination of membrane proteins, we have recently started "ERATO human receptor crystallography project" supported by the Japanese Science and Technology Agency. We have also obtained a support from the Wellcome Trust to establish an outstation of Imperial College London at the new UK synchrotron Diamond. I will discuss our strategy how to establish the structure determination method of human membrane proteins using these new facilities and its impact on biological sciences, pharmacology and medicine.

View full abstract

In hit-finding processes, more hits and much structure-activity relationship information are required to be extracted through biological assays. For this purpose, we developed the active learning method, which captures many aspects of chemical information. To evaluate our methods, we used the structure-activity data of the compounds which bind to GPCRs (G-protein coupled receptors). In this study, we have used 1,551 compounds which bind to biogenic amine receptors selected from Pharmaprojects database as positives, and 255,440 compounds selected from Pharmaprojects and Available Chemicals Directory reagent database as negatives. We have shown that the active learning is efficient for hit-finding and prediction. Based on our method, we have shown that 11% of biochemical experiments of the whole chemical library are enough for finding 90% of positives, and 32% are enough for finding 99% of positives. On the contrary, for finding 90% (or 99%) of positives, 90% (or 99 %) of biochemical experiments were necessary when random screening methods were conducted, and 22% (or 61%) of biochemical experiments were necessary when similarity-based screening methods were conducted. Using the active learning method, we developed rapid and cost-effective systems for hit-finding.

View full abstract

Although high throughput screening (HTS) of millions of compounds in the early stage of drug discovery has become very common, it is not necessarily cost effective. Therefore, designing focused libraries targeting certain proteins is recently receiving more attention. PharmaDesign virtually screens millions of compounds against 3-D structures of target proteins in silico, and designs focused libraries. We designed focused libraries for chemokine receptors, a family of G-protein coupled receptors (GPCRs), which are known to have very limited structure information available. When an inhibitory assay was conducted against one of the chemokine receptors, CCR4, the hit rate was 13.3%. Furthermore, we designed focused libraries for DAP(Death Associated Protein) family, which regulates apoptosis/autophagy. We believe that focused libraries designed based on 3-D structures of target proteins significantly contribute to more efficient identification of hit compounds, and accelerate drug discovery process.

View full abstract

We have started to develop Computed Property Data Base: CPDB. A part of the CPDB has been opened to the public at RIO-DB (Research Information Database) of the National Institute of Advanced Industrial Science and Technology, Tsukuba from January 2005 (http://www.aist.go.jp/RIODB/hrmsdb/index.html). Thirteen molecular properties of 258,907 molecules were stored in CPDB. All of the properties are estimated using mainly the Joback method which is implemented in an Integrated Software Tool for Molecular Design: MolWorks. It is possible to retrieve not only molecular property of input but also molecules which have inputted properties. CPDB is an attempt to develop a system for efficient molecular screening using estimated molecular property.

View full abstract

The most important issue for drug discovery is to have an enormous library of high quality compounds, that can provide several potential hit compounds of various scaffold types, all of which should be suitable for lead optimization in the lead-generation process. To enlarge the corporate compound library by adding compounds from millions of vendor catalogue compounds, a compound has to satisfy the following two requirements: one is that the compound is inside the drug-like space targeted in the drug discovery study and another is that it is located in a diversity hole where few in-house library compounds are already available. In this study, a cheminformatics method to quantify chemical space has now been developed using a 14-dimensional principal component analysis space and a simple equation has been proposed to define the absolute drug-like space.

View full abstract

For the chemical file enhancement, it is quite important to acquire good compounds for drug discovery. One of the approaches to find the GOOD compounds is the visual inspection by the well-experienced medicinal chemists. Recently, due to drastic increase of the commercially available compounds, it has become impractical to inspect all of them solely by their eyes and knowledge. For last few years, we have introduced some computational filtration protocols, such as Lipinski Rule filter, Bayesian model-derived Drug-likeness Score filter, etc. They are employed for preprocessing steps in the inspection so that the medicinal chemists can skip unfavorable compounds automatically. By means of the filtration protocols, our inspection of compounds has become more exhaustive and efficient. Because of the size of the acquisition, its cost performance is another point of interest. In the sense of improving the cost performance, it is preferable to acquire compounds from one well-qualified supplier. A set of the filtration protocols can be employed as a tool for such qualification as well. In this presentation, the speaker will talk about some examples of these approaches.

View full abstract

We have constructed a natural products database, "RIKEN Natural Products Encyclopedia (RIKEN NPEdia)", as an informatics part of chemical resource bank, "the RIKEN Natural Products Depository (RIKEN NPDepo)". Approximately 25,000 compound data are now listed in NPEdia, some of which were transferred from "KNApSAcK", a secondary metabolites database constructed by Prof. Kanaya and his colleagues (http://kanaya.aist-nara.ac.jp/KNApSAcK/, http://prime.psc.riken.jp/ (mirror site)). Compounds listed in NPEdia are mainly secondary metabolites isolated from actinomycetes, fungi, plants and other organisms. Each compound data consists of molecular structure, origin, and physico-chemical and biological properties. MOL files can be also downloaded. Ultra-Violet, MS/MS and NMR spectra will also be included in the near future. This database involves three functions: (1) searching for compounds, (2) registering compound data and (3) co-operating in compound-order system. We are now trying to integrate metabolites with metabolic pathway maps, protein-protein interaction maps, and chemical genomics maps (relationship between metabolic pathways and small molecules "bioprobes" interacting them). We think NPEdia should be a useful database friendly to researchers in various research fields. The details and latest news about NPEdia are described on the website (http://npd.riken.jp/).

View full abstract

High-resolutional mass spectrometry with a combination of HPLC or CE are widely applied to life sciences. When extracts from cells or tissues are analyzed with such LC-CID-high-resokutional MS, they give several thousands of peaks. However, only a few fraction of the spectral data is manually analyzed to identify chemical sturctures of metabolites; the rest of the spectra is left unidentified. Development of a tool to interprete high-resolutional mass spectra automatically and to help to elucidate chemical structures is essential for metabolomics. For the development, it is essential to construct a database by collecting high-resolutional mass spectra. From April 2006, we have started a project to construct such a database and to develop tools for helping automatic structure elucidation. Here we report about the goal, strategy and present status of Metabolome Mass Spectral Database that has started from April 2006.

View full abstract

Secondary metabolites are highly species specific and play a role for the survival of the producing organism within its natural habitat. Several databases have been made by collecting metabolite information of various organisms, and provide some chemical information and biological pathways on metabolites, however, they don't provide the relationships between metabolites and their biological origins. To systematically and comprehensively understand species-specific diversity of metabolites, we have designed a database system called KNApSAcK. This system is useful for obtaining information on metabolites and their corresponding species, chemical structure and biological activity. The database has a tool that can be used for analyzing datasets acquired using Fourier transform ion cyclotron mass spectrometry (FT/ICR-MS). We collected information on 25,930 metabolite-species pairs encompassing 11,075 metabolites and 8,557 species from published references (January 19th, 2006). This database system and online manual are freely available at http://kanaya.naist.jp/KNApSAcK/. This database system is available in two different versions, Web-version and Download-version. If you want to use the Download-version, at the beginning, you have to install KNApSAcK and Java 1.4.2 on your local computer. The Download-version has been coded in Java.

View full abstract

We have been investigated the MO/MC method, Monte Carlo simulation using Molecular Orbital calculations. This method requires enormously long CPU time because we have to execute MO calculations more than 100,000 times to obtain statistical information. We have developed a new algorithm to parallelize the MO/MC method. This algorithm is follows. i) Input an initial geometry and calculate its energy. ii) Generate the next ones and calculate energies on multiple machines. iii) Calculate differences between the energy on each machine and the previous one, ΔE. iv) Select a machine with ΔE according to a random number, then v) determine the translation to the new geometry on the selected machine to be allowed or not under the defined conditions. vi) If the translation was not allowed, select another machine using a random number and repeat iv) and v), and otherwise, return to i). This algorithm is expected to shorten CPU time less than half in using 4CPUs in our estimation. It took ~1370k seconds on the Intel(R) Pentium(R) D 3.2GHz machine and ~670k seconds (8days) at the 4 CPU one to execute a MO/MC calculation of 2,000,000 steps of a 121 atoms cluster using the PM3 level of theory. The ratio 0.489 as expected.

View full abstract

It is possible to use theoretical calculations in predicting whether or not toxic substances easily decomposed to others in the environmental conditions. It is because calculated activation energies Ea(calc), are closely related to the rates of decomposition of substrates or observed activation energies, Ea(obs). In order to confirm this concept, the acid-catalyzed hydrolysis of esters whose mechanism has been already confirmed was investigated and their activation energies, Ea(calc), were calculated for 13 esters at the 6-31G* level of theory. We obtained weak correlation between Ea(calc) and Ea(obs) (R2=0.7655). According to the detailed analysis of geometry of reactant for methyl methacrylate and methylacrylate, it was confirmed that the reactant geometry of the former ester, especially positions of water molecules, is very much different from that of the latter. In order to avoid errors coming from the differences in the geometry of reactants, we tried to correlate Ea(obs) with E(dif), the energy difference between the total energy of the transition state (TS) and those of ester as well as two water molecules. We obtained a better correlation with R2=0.8598. It is considered that the correlation between the two energies can be used to predict how easily esters decompose to others.

View full abstract

We developed a new parallel density-functional canonical molecular-orbital program for large molecules based on the resolution of the identity method. In this study, all huge matrices were decomposed and saved to the distributed local memory. The routines of the analytical molecular integrals and numerical integrals of the exchange-correlation terms were parallelized using the single program multiple data method. A conventional linear algebra matrix library, ScaLAPACK, was used for matrix operations, such as diagonalization, multiplication, and inversion. Anderson's mixing method was adopted to accelerate the SCF convergence. Using this program, we calculated the canonical wavefunctions of a 306-residue protein, insulin hexamer (26,790 orbitals) by the direct-SCF method. The total parallelization efficiency of the first SCF iteration was estimated to be 82% using 64 Itanium 2 processors connected at 3.2 GB/s (SGI Altix3700), and the calculation successfully converged at the 17-th SCF iteration. The whole computational time including the calculation before the SCF loop was 2 days and 17 hours. This study put the calculations of the canonical wavefunction of 30,000 orbitals to practical use.

View full abstract

CAST/CNMR, a system for 13C NMR chemical shift prediction and structure elucidation, is designed for practical use by highly experienced structure analysts. We provide highly accurate prediction based on careful deliberation on significant factors in the NMR prediction, including stereochemistry, from the viewpoint of analysis of complicated structures. The CAST/CNMR system has been actually applied to several practical problems in structural analyses of natural products and their synthetic intermediates. This paper describes a part of the several applications concerning correction of chemical shift assignments and structural revision of natural products.

View full abstract

The molecular orbital is a fundamental quantity to describe electron's behavior. The number of nodal regions is often used as an index for characterizing molecular orbitals. We have been developing a visualization system MOOTIC which enables observation of the metamorphosis of molecular orbital according to the molecular frame distortion in real time. The metamorphosis of iso-surface of molecular orbital is also observed. In our system, a fast rendering method is proposed to display the iso-surface by interpolating a pair of voxel data sets similarly to the keyframe animation, that is an image generation method to blend a pair of key frame images into the interpolated frame. In advance, a series of voxel data sets is constructed by changing the molecular frame position, called the key position. In Windows platform, MOOTIC succeeded to visualize the virtual orbital of hydrogen molecule, the carbon monoxide(CO), and the protnated Schiff base of retinal(PSBR). Changes of molecular orbitals depending on bond elongation for CO and bond rotation for PSBR are also visualized.

View full abstract

We developed programs for classifying membrane proteins to seven major categories, using nearest neighbor method (NN) and self-organizing map (SOM). The same representation of protein amino acid sequences was employed as the one for GPCR sequences, for which we developed classifying programs with very high performance. That is, we used vector representation obtained by auto-cross covariance derived from z-scales that were computed as principal components of 26 physicochemical properties of amino acids. Actually, we constructed input vectors by concatenating those auto-cross covariance over several lags. We used 115,447 membrane protein sequences extracted from non-redundant database of NCBI. The seven major classification categories are transporters, ion channels, receptors, membrane enzyme, cytadherence proteins, cytoskeletal protein, and facilitators. We applied leave-one-out cross validation to NN, and four-fold cross validation to SOM. As a result, NN gave precision of 91.90%, and SOM 90.52% for training data and 88.99% for test data. The precisions looks fairly good considering rather rough preprocessing of protein data and gives prospect to improve through refinement of training data.

View full abstract

Diagnostic nuclear medicine image including SPECT enables us to see the fluctuation in blood flow. Thus, the brain imaging by SPECT is used for the diagnosis of diseases such as Alzheimer's disease, which reduces regional cerebral blood flow (rCBF). Although the diagnosis has been made visually by the expert reader conventionally, some problems, such that subjectivity can not be completely excluded, have been pointed out. Recently, from the EBM (evidence based medicine) perspective the statistical evaluation methods for diagnostic image have been widely studied in order to diagnose as objectively as possible.

View full abstract

Halogenated aromatic hydrocarbons(HAHx) and polycyclic aromatic hydrocarbons(PAHs) cause serious problems as environmental contaminants. They develop various adverse effects including body weight loss, cancer promotion, immunosuppression through the aryl hydrocarbon receptor(AhR). And they are often taken by mouth. Anthraquinones, one of the plant pigments and exist in some vegetables, suppress AhR transformation induced by HAHx and PAHs. Therefore it is expected that the anthraquinones will suppress the adverse effect by HAHs and PAHs taken together. We analyze the relationship between their antagonist activity to AhR and descriptors by using the Partial Least Square(PLS) to estimate the defunct anthraquinones activity and to extract the physical and constructional features which cause the activity. We divide the descriptors to 2 groups (physical properties descriptors and construction descriptors) and apply the PLS to each group, then pick up the descriptors which are highly contributed to make the regression model from each result and apply the PLS again. This method is based on the theory of covariance structure analysis. We could obtain better model compared with the model constructed by all descriptors. And the constructional features, especially the position of hydroxyl function are deeply related to the degree of the antagonist activity.

View full abstract

We propose a novel regression method based on support vector regression (SVR), which is an expansion of support vector machine to construct nonlinear regression model. In these types of regression methodology, variable selection is a one of the most important issue to construct robust and predictive QSAR model. Thus, we adopted a variable selection method, which is based on sensitivity analysis of each variable, and compared with a GA-based variable selection method. A usefulness of the proposed method has been verified by applying to real QSAR data set, benzamidine derivatives as inhibitors of trypsin and thrombin. Steric and electrostatic interaction energies were calculated between the derivatives and each amino acid residue of receptors, and were used as chemical descriptor. As a result of construction of SVR models, predictive QSAR models were obtained for both receptors. Moreover, variable selection procedure based on sensitivity analysis was successfully determined a set of important variables to explain inhibitory activities for trypsin. The R2 and Q2 values of the final SVR model containing twenty-five significant variables are 0.857 and 0.619, respectively.

View full abstract

We have developed a novel catalyst design method using multi-objective optimization and evolutionary computation. Relationships between objective variables (conversion and selectivity) and explanatory variables (catalyst compositions, temperature, pressure, etc.) are modeled by using PLS (partial least squares) method. Explanatory variables are selected by use of GAPLS (genetic algorithm based PLS) method in order to improve predictive ability of the models. Then, these models are used for multi-objective optimization of conversion and selectivity by use of evolutionary computation method. In this study we adopted NSGA-II (nondominated sorting genetic algorithm II) as a multi-objective optimization algorithm because of the high efficiency of computation. Pareto-optimal solutions acquired by the multi-objective optimization provide compositions of catalyst that are expected to be suitable for required catalytic performance. We collected catalytic data about direct epoxidation of propylene from patents and have applied the approach to this data. The results showed that our approach is effective in catalyst design.

View full abstract

In this study, Piecewise Kernel Partial Least Squares with Bagging k-means (PKPLS) is proposed for more effective QSAR analysis. PKPLS is based on Kernel PLS and piecewise modeling with bagged k-means. Piecewise modeling aims to divide samples into several clusters and apply PLS modeling to each cluster. In this piecewise modeling, k-means is used for clustering. K-means is simple algorithm and it requires low computational cost but the problem is that it strongly depends on initial value randomly selected. To improve its stability, bagging is introduced to k-means, namely, k-means is performed several times with different initial values and final result is made as an ensemble of all result. In PKPLS, the input data are projected to higher dimensional space by a kernel function. Then, piecewise modeling with bagging k-means is applied to the projected data. By combining piecewise modeling and KPLS, PKPLS has both advantages of these methods. Namely, it can deal with nonlinearly and any kinds of sample distribution. The potential of PKPLS is demonstrated with an QSAR data set of ngiotensin II Antagonists. The result indicates that PKPLS shows better regression performance than that of PLS.

View full abstract

In order to minimize the risks to human health and the environment, it is required the identification of potentially hazardous chemicals from more than 20,000 untested chemicals under the Japanese Chemical Substance Control Law. Our organization made plans to evaluate the biodegradability and bioaccumulation of untested chemicals by using structure-activity relationship to identify persistent and bioaccumulative chemicals. In previous studies, we validated biodegradability prediction models and bioconcentration prediction models. In this study, we constructed batteries of three validated biodegradation prediction models (CATABOL, CERI model and BIOWIN) and three validated bio-concentration prediction models (Baseline Model, CERI model and BCFWIN) based on Bayesian analysis. Biodegradation data for 1321 chemicals and Bio-concentration data for 808 chemicals were as the validation set. The accuracy of both predictions for biodegradation and bio-concentration by the batteries of models were better than that by individual model. We found that the classification of prediction results by dividing calculated values into small segments are effective for getting useful information for further analysis.

View full abstract

Fujita's proligand method (Fujita S (2005) Theor. Chem. Acc. 113:73-79, 113:80-86, Fujita S (2006) Theor. Chem. Acc. 115:37-53) is applied to the enumeration of three-dimensional trees as stereochemical models of alkanes. Such 3D-trees are classified into centroidal and bicentroidal 3D-trees, which correspond to respective promolecules having proligands as substituents. The generating functions of planted 3D-trees are introduced into functional equations obtained. Thereby, the numbers of 3D-trees or equivalently those of alkanes as stereoisomers are calculated and collected up to carbon content 20 in a tabular form. Now, the enumeration problem initiated by Cayley (Cayley A (1874) Philos. Mag. 47(4):444-446) has been solved in such a systematic and integrated manner as satisfying both mathematical and chemical requirements.

View full abstract

The topological index Z was proposed by the present author in 1971 for characterizing the topological structure of the carbon atom skeleton of hydrocarbon molecules, which was found to have good correlation with the boiling point of alkanes. The Z index thus has been used in thermodynamic QSAR study and later in developing the "graph-theoretical molecular orbital theory." However, recently Z was found to play a very important role in elementary mathematics and algebraic number theory by transforming algebra into geometry or graph-theory. Examples are Fibonacci and Lucas numbers, Pascal's triangle, Pell number and Pell equation, Pythagorean and Heronian triangles, etc. In this paper these mathematical features of the Z index are introduced and discussed.

View full abstract

There have been many reports on the capillary gas chromatograph column retention times (GCRT) for various environmental contaminants, because it is well known that they are one of the useful parameters for predicting the environmental fate, and/or many physicochemical properties such as aqueous solubility, n-octanol/water partition coefficient, and Henry's low constant for the contaminants. However, it seems that a detail analysis from a thermodynamic point of view is difficult, because they are not only observed under the programmed temperature condition, but also they closely depend on the change in the column temperature. In the present study, the GCRT under isothermal conditions were observed for tetrachlorobiphenyl isomers known as very serious environmental contaminants. It was found that there was a marked difference between the GCRTs observed under (1) the programmed condition and (2) the isothermal temperature at 185 deg. C condition. A relationship between the GCRT observed under the isothermal condition and the second order connectivity indices defined by Randic (2c) was investigated. A linear relationship between them was found in terms of the classification on the basis of the number of the pair of chlorine substituents at the proximitive position.

View full abstract

XyMML(XyM Markup Language) has already been developed as a basic technology to display and to transmit a chemical structural formula with WWW(Worls Wide Web). A chemical structural formula can be displayed immediately on a browser by translating a XyMML code into a XyM Notation, which is interpreted by the XyMJava system. As a result, this can be applied as a system for searching chemical structures.

View full abstract

ChemoJun, a new open source graphics library for chemical visualization, has been developed and opened to the public from 26 December 2005. The purpose of the ChemoJun project is to develop a basic technology of chemical software. ChemoJun consists of several basic modules for visualizing chemical information. These modules are used in developing graphical user-interface programs of chemical software, including FRAU, CAST/CNMR, and HaptiChem systems. The usage statistics of the files downloaded from the ChemoJun site (http://research.nii.ac.jp/~cheminfo/ChemoJun/index-e.html) show that the files have been constantly downloaded, and the total number of 44,512 files and an average of about 180 files per day were down loaded from 26 December 2005 to 31 August 2006. This paper describes the outline of ChemoJun and the latest version, 050, which has been released on 10 August 2006.

View full abstract

The maximum complete subgraph is used for many kinds of problems, e.g., to find similar structural chemical compounds, to search similar proteins and to solve docking problem of protein and substrate. We also used the subgraph for detection of active sites on protein structures. To find a maximum complete subgraph from a given graph say the maximum clique problem. As the maximum clique problem is NP-hard problem, there has been prepared some fast solved algorithms. The algorithm by Carraghan and Pardalos is popular one and we used this algorithm to our protein active sites detection system. Recently another fast algorithm was presented by Ostergard. To clear the characters of those algorithms, we prepared some kinds of graph and systems implemented those algorithms including some slightly revised versions and measured computational times for solve the maximum clique problem. The results indicate that the algorithm of Ostergard is useful in graph with small number of nodes and low edge density. In our protein active sites detection system, it uses a graph with much number of node indicating pairs of amino acid residues of two proteins. Thus we evaluate that the Carraghan's algorithm is adapted for our system.

View full abstract

Recently, Web technologies have been making remarkable progress. Web User Interface (WUI) provides easy operating environment and lets us access to a remote resource without knowledge of troublesome handling. These advantages are quit suitable for parallel and distributed systems, because it is generally difficult for non-specialists to build up and use these systems practically. In fact, several Web portal sites use these technologies and contribute to provide easy interface of parallel and distributed environment to ordinary user. We have also developed our portal site for molecular computational tools (CONFLEX.net), and performed some practical experiments via internet. The site provides simple and easy interface to compute molecules without knowledge of computational resource environments. It has not only supported a single application (CONFLEX), but also can introduce several applications (Gaussian, GAMESS) as a recent improvement. In this conference, we will talk about the other improvements; new protocols for data communication among distributed computer components, control of application-level parallel and distributed computation, management of user limitation for available computational resources, etc.

View full abstract

Substituent effect on the acidities of phenols in water is considerably different from that in gas phase. That is, acidities of phenols which have electron-withdrawing groups at the para position are intensified in comparison with that in gas phase. In order to make clear the origin of this phenomenon, ab initio calculations were done. Energies of substituted phenoxides and corresponding phenols were calculated to determine relative acidities in gas phase for phenols. In addition, energies of corresponding species, which are specifically hydrated in some manner, are also calculated. Some kinds of relative acidities of phenols, which include specifically hydrated components, were determined. Their substituent effects were compared with that in gas phase and experimental one in aqueous phase. It was developed that the specific solvation to the ring substituents of phenoxides by one water molecule is the intrinsic factor to reproduce the acidities of phenols in water media.

View full abstract

Intermolecular interactions play a significant role in the field of physics, chemistry and biology. The study of intermolecular interactions is a useful step toward gaining a basic understanding of the dynamic behaviour of molecules in various environments. In the present study, high-level ab initio calculations [MP2] using correlation consistent basis sets of Dunning [aug-cc-pVDZ, aug-cc-pVTZ] were carried out to study the potential energy surfaces of C2H6/H2O and C2H5F/H2O Van der Waals dimers and to clarify the stable structures of these clusters. For the calculation of interaction energies, the basis set superposition errors (BSSE) were estimated by counterpoise method. One stable structure has been found for the C2H6/H2O dimer. The optimized structure of the C2H6/H2O has the H2O molecule in an equatorial location of the C-C axis. For the C2H5F /H2O dimer, two stable structures are found. One has the H2O molecule in an equatorial location of the C-C axis, and the other has the H2O molecule in an axial region of the C-C axis. The details of the potential energy surfaces are discussed.

View full abstract

A Cycl[3.2.2]azine molecule ([2.2.3]cyclazine) has a rigid C2v plane structure with a sp2-hybridized carbon perimeter clipped by a central nitrogen atom, as an isoelectronic analogue of a [10]-annulene. Various types of cyclazines have been reported and attracted much attention from physicochemical, magnetic, biological as well as graph-theoretical viewpoints. Despite their interesting properties, no systematic quantum chemical analysis have been reported to characterize the UV/Vis intense electronic peaks and the associated sidebands. In this study, semi-quantitative (TDDFT) and quantitative (MS-CASPT2) analysis are presented for a cycl[3.2.2]azine, its benzo-annulated and the maleimide-linked derivatives. Furthermore, the detailed spectral structures were analyzed on the basis of the orbital vibronic coupling interaction scheme.

View full abstract

Ab initio molecular orbital calculations were performed to clarify the reaction mechanism of the repair process by photolyase of thymine dimer (6-4) photoproduct, using HF/6-31G* and MP2/6-31G* levels of theory. Regarding an anion radical of (6-4) photoproduct as a reactant, we optimized the geometry of the intermediate and the transition state. We found that one of the bonds of four-membered ring breaks easily when thymine dimer gets an electron. To design an inhibitor which makes the reaction of photolyase later, we also calculated 7 kinds of analogues in which one or two oxygen atoms other than the oxygen atom that composes four-membered ring of oxetane are replaced by sulfur atom, and compared the reaction energies of the analogues. We found that the activation energy increases when sulfur atoms are substituted for oxygen atoms. This suggests that S-containing thymine dimer is likely to be an inhibitor for photolyase.

View full abstract

Density functional calculations have been employed to study the reaction of triplet Pb with NO2 at the B3LYP/aug-cc-pVTZ-PP level. Association of Pb and NO2 and subsequent dissociation to PbO and NO are considered. Structural parameters, vibrational frequencies and relative energies are calculated for the associated species, complexes and transition state. PbNO2, trans-PbONO and cyclic PbONO are barrierlessly obtained as stable associated species. The formation of cyclic PbONO is calculated to be most exothermic. In the fragmentation path to singlet PbO and NO two van der Waals complexes are found to exist. The reaction on the quartet state will be presented briefly.

View full abstract

Theoretical studies on ks-kD rate ratio in the solvolysis of 2-phenylethyl tosylate was carried out by using ab initio MO calculation (RHF/6-31G* or B3LYP/6-31G*) combined with polarizable continuum medium model solvation energy evaluation. Geometry optimization of each reaction pathway in the gas phase afforded reaction activation energy of 6.22 kcal/mol (RHF) or 7.25 kcal/mol for kD, and 8.85 kcal/mol (RHF) or 12.35 kcal/mol (B3LYP) for ks pathways. The results were inconsistent with the experimental ks/kD rate ratio of 30 in the solvolysis of 2-phenylethyl tosylate at 75 oC in water. The theoretical reaction activation energies in water were 14.74 kcal/mol (RHF) or 14.69 kcal/mol (B3LYP) for kD pathway, and 12.51 kcal/mol (RHF) or 11.11 kcal/mol (B3LYP) for ks pathways. The theoretical results in water indicate the preference of ks over kD pathway in the reaction, which is consistent with the experimental results.

View full abstract

We have synthesized four hetero cyclic compounds of 1-(S)-N-[2-(diphenylphosphinyl) naphthyl-2-(hydroxylmethyl) pyrrolidine (1), 1-(S)-N-[2-(diphenylphosphinyl)naphthyl] -2-(pyrrolidinyl-methyl pyrroridine (2), 2-(diphenylphosphanyl)-N, N-dimethyl-1-benzamide (3) and 2-(diphenylphosphanyl)-phenyl-pyrrolidin-1-yl-methanone (4), and investigated their mobility in solution by NMR. Conformational analysis of the heterocyclic compounds is useful in deducing the structure in which they are active as a catalyst. In the present NMR measurements, 1H-X fg-JHMBC (field gradient J-Resolved Hetero-nuclear Multiple-Bond Correlation) spectroscopy was implemented as a tool for the determination of hetero-nuclear three bond, phosphorus and protons and carbon-protons coupling constants. By fitting a sine curve to the experimental data by the method of 3D J-resolved HMBC NMR measurements, accurate nJHX coupling constants were obtained. From the coupling constants, the corresponding dihedral angles, H3-C2-C3-P, H12-C12-C11-P, H16-C16-C11-P, H18-C18-C17-P, and H22-C22-C17-P, of these hetero cyclic compounds were determined. We carried out optimization of the four compounds using the PM3 and ab initio methods with and without the aid of NMR data for the dihedral angles. The Karplus formula for the relationship between dihedral angles and coupling constants were confirmed to be effective in conformational analysis by 3DNMR. The most stable structure of the compounds was the calculated one in which the dihedral angles experimentally determined were adopted.

View full abstract

Recently the constrain geometric catalyst (CGC) have become one of the most important catalyst systems for the Ziegler-Natta polymerizations of olefins. In this conference, the homogeneous Ziegler-Natta reaction mechanisms of olefin with CGC by incorporation of eight bridging atoms or groups (BH, CH2, NH, O, AlH, SiH2, OH, and S) at the cyclopentadienyl ring and amido group as ligands are reported. These potential energy surfaces were calculated by a density functional method as a B3LYP. The first step of the reaction is the complex formation with stabilization energy (from 23.5 kcal/mol to 27.1 kcal/mol) and the second is the olefin insertion process through the transition state. The energy barriers at the transition states from the reactants for all systems treated here indicate negative values. Accordingly the rate-determinant step of olefin insertion is probably the complex formation process. The formation energy of the complex between ethylene and CGC by incorporation of bridging of BH group is the most stable, and the second stable complexation energy is produced by incorporation of bridging of O atom. It was clarified the role of the bridged atom or group to the potential energy.

View full abstract

Fluorescence of naphthanthrone, 6H-benzo[cd]pyren-6-one, in degassed organic solution increases on light irradiation, but it stays almost constant in aerated solution. To clarify this phenomenon, the absorption and emission spectra of naphthanthrone were measured in several solvents as a function of irradiation time at room temperature and 77 K. The fluorescence enhancement was observed only in solvents that contain hydrogen, such as methanol and benzene, and in ethanol even at 77 K. The enhancement occurred after several minutes irradiation, continued to grow with irradiation up to 30 minutes, and then began to reduce; at 77 K the enhancement was saturated after 2 h irradiation and no reduction occurred. On the basis of the experimental data we proposed the mechanism of the fluorescence enhancement. Naphthanthrone excited to the triplet state on irradiation tends to abstract hydrogen from the solvent, but fails to complete the abstraction because of its low energy. Instead it forms an association compound with a solvent molecule and the compound provides strong fluorescence near the wavelength region of the fluorescence of naphthanthrone. The absorption spectra calculated for the compound by the ZINDO method were consistent with the experimental results.

View full abstract

The log of the equilibrium constant for partitioning a solute between octanol and water (logP) is a leading term in many quantitative structure-activity relationships (QSAR). The logP can be evaluated by the difference of solvation energies between in water and in octanol. The continuum solvation model have been widely used for the estimation of solvation energy in water. In this study, we coded the GB/SA calculation using Merck Molecular Force Field (MMFF94s) charges and van der Waals radii. And the parameter sets for water and n-ocatnol are optimized to reproduce the experimental logP values. We will discuss the accuracy and usefulness of the logP calculations using GB/SA model with MMFF94s.

View full abstract

It is well known that frontier orbitals of molecules are important to understand chemical reactivity. We have already published papers on relations between frontier electron densities and toxicities of dioxins and dibenzofurans. In this work, we investigate the relations between frontier electron densities and activities of polyphenols. Highest occupied molecular orbitals (HOMO) and lowest unoccupied molecular orbitals (LUMO) of 15 polyphenols are calculated by semi-empirical AM1 and ab initio Hartree-Fock molecular orbital method. Electron densities of LUMO of 15 polyphenols are used as variables for multivariate statistical analysis. 15 polyphenols are classified according to their principal components.

View full abstract

Molecular structures for dimers of enol forms of cyclic conjugated 1,2-diketones, tropolone (1), 2-hydroxycyclopentadienone (2), and 2-hydroxycycloprpenone (3), in simultaneous intermolecular double proton transfer were investigated by DFT (B3LYP/cc-pVTZ) calculations. Although two conjugated carbon rings in the (1) and (2) dimers were essentially coplanar in the ground state, they were severely twisted to each other in the transition state. The two cyclopropenone rings of the (3) dimer were, on the other hand, coplanar both in the ground and transition states. From close examination of the hypothetical coplanar structures in the transition state for the (1) and (2) dimers the twisted transition-state structures were ascribed to the short O...O hydrogen bonds with almost linear arrangements of the intermolecular O...H...O hydrogen-bonding atoms. Large bond-angle strains in the COH angles in the hypothetical coplanar transition states could be relieved with significant torsions of the C-O...O-C dihedral angles, resulting in the severe twisting of the two conjugated carbon rings. The COH angles for the (3) dimer were nearly constant both in the ground and transition states, and its molecular structure are therefore coplanar during the proton transfer.

View full abstract

In the structure-based drug design, it is important to know the binding geometry of the ligand to the protein. Because, by knowing it, we will be able to develop the strategy for modifying the structure of the ligand to enhance the inhibitory activity. In case that the co-crystalline structure is unknown, we have to estimate the accurate binding geometry. For this purpose, a lot of software has been developed, but the accuracy is not satisfactory. Here, we report the new method for estimating the binding geometry. We also considered flexibility of the ligand and the protein in the estimation. We confirmed that the new estimation method in which rigid-body molecular mechanics calculation is used can reproduce the binding geometry of the observed co-crystalline structure. Further, we confirmed that we can improve the accuracy in binding geometry estimation by taking into account the flexibility based on the normal mode analysis.

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 DNA. In this study, we inspects applicability of FMO method to ab initio Molecular Dynamics (MD) method by verifying the accuracy of Potential Energy Surface(PES) of a catechine molecule between FMO method and conventional method. The accuracy of PES is 1.1kcal/mol within the range of 1eV from the energy minimum. Therefore we think that FMO method is applicable to ab initio MD.

View full abstract

In our continuing efforts to simulate biological compounds, the quantum chemical calculations on the complexes of HIV-1 protease with its inhibitors were carried out by employing ab initio fragment molecular orbital (FMO) method. The PDB X-ray crystallographic structures of HIV-1 protease in complex with six different inhibitors, indinavir, lopinavir, ritonavir, saquinavir, SD146, and XK216, were optimized by energy minimization using OPLS-AA force field implemented in MacroModel, respectively. These six inhibitor-bound HIV-1 protease complexes were subjected to a FMO calculation at MP2/STO-3G level by using BioStation package. The FMO calculations showed that the binding energy between SD146, a cyclic urea HIV-1 inhibitor, and HIV-1 protease is larger than those between other inhibitors and the protease. In addition, we found that the interaction of two cyclic urea HIV-1 inhibitors, SD146 and XK216, with Asp25 residues of both α- and β-chains in the HIV protease is large compared to that of four peptidomimetics HIV-1 inhibitors with Asp25 residues. Finally, it is anticipated that the distinct binding and interaction energies can be related to the side effects of HIV-1 inhibitors and their potencies against mutant strains of HIV.

View full abstract

"NMR-2JCH index" analysis developed by Oikawa and co-workers is a simple and reliable technique for identification of various saccharides, and the improved application toward elucidation for the primary sequence and three-dimensional structure of biologically important oligosaccharides is strongly expected. In the latest extension of the 2JCH index analysis to various aminosugars and their N-acetyl derivatives, we have observed that only 2JC4,H5 value of alpha-galactosamin (a-GalN) is peculiarly different from those of the others. In order to explain the problem, we have performed various computational conformational analyses of a-GalN and the N-acetyl derivative (a-GalNAc) by using CONFLEX conformational search method. In this study, various theoretical levels with solvent effects have been also examined for exact predictions of NMR coupling constants on the pyranosyl ring of 20 a-GalN conformers found by CONFLEX. Finally, it is noted that the calculated 2JCH values of theromodynamical conformational mixture based on the Bolzmann law using conformational energies and 2JCH values at b3pw91/6-311+G(2d,p)// b3pw91/6-31+G** level calculation in water, can represent the experimental values within 1.3 Hz RMSD. We are now investigating the origin of the remarkable differences of 2JCH values between a-GalN and a-GalNAc.

View full abstract

We analyzed geometrical isotope effect (GIE) and kinetic isotope effect (KIE) in R-H(D)···R type proton(H)/deuteron(D) transfer on intramolecular hydrogen bond of malonaldehyde (R=O) and its isoelectronic compounds: 1,5-diaza-penta-1,3-diene (R=NH) and penta-1,3-diene (R=CH2) using Hartree-Fock level of theory of multi-component molecular orbital (MC_MO-HF) method to treat quantum effect of nuclei explicitly. Single s-type GTF are used for each proton and deuteron with 6-31G** electronic basis set. The bond lengths of R-H(D) and H(D)···R in both equilibrium and transition state are the order of C, N, and O due to the strength of hydrogen bond (O>N>C). By delocalization of nuclei, the bond length of R-H(D) and H(D)···R derived from MC_MO method are longer and shorter than those of conventional MO in both case of H and D, respectively. As GIE, the distances of R-D and D···R are longer and shorter than those of R-H and H···R, respectively. The activation energies by conventional MO are larger than those by MC_MO method. Because the ratio of rate constant of penta -1,3-diene as KIE is 4.03 in conventional MO and 7.53 in MC_MO method, Result of MC_MO method shows good agreement with observation value, 12.2.

View full abstract

In the situation that environmental regulation is strengthened year by year, the method and catalyst development of removal NOx from exhaust gases that can remove NOx effectively are demanded. In late years selective catalytic reduction of the NOx by hydrocarbons is an effective process in the presence of excess oxygen. In this study, We used Ag/gamma-Al2O3 with high selectivity in a catalyst used for this reaction and We investigated reaction mechanism of NO oxidation reaction on Ag/gamma-Al2O3 catalyst by ab initio calculation. We calculated Z direction optimization only O atom which maintained adsorption species and reaction site Ag and fixed the other atomic distance and angles in documents value. In only catalyst, the O atom and Ag atom acted as electron acceptor together, and the Al atom as electron donor. Stabilization of energy was observed by O2 adsorption. It is thought that charge movement from adsorption site Ag to adsorption species O2 that is contribution of Back Donation influenced this.

View full abstract

A new Partial Least Squares (PLS) method, Ensemble PLS (EPLS), based on bagging and boosting is proposed in this study. Empirically, it is known that boosting improves classification or regression model more than bagging, however, bagging has superior robustness to the noise compared to boosting. EPLS takes both advantages of boosting and bagging by combing bagging PLS and boosting PLS. Bagging PLS and boosting PLS are based on the ensembles of numerical PLS models built with a part of the sample. The difference between bagging PLS and boosting PLS is their ways of resampling. In bagging each regression model is built with a part of the samples randomly picked up. On the other hand, in boosting, samples which the previous model cannot predict well are purposely picked up. In EPLS, first, several sets of samples, called subset, are picked up randomly and boosting PLS is applied for each subset. Predictions of all subset are combined by bagging PLS. The performance of EPLS is demonstrated by a QSAR data set of Angiotensin II Antagonists. The result clearly reveals that EPLS yields superior prediction results to PLS.

View full abstract

In this study, I calculated electronic intelligence from a ligand coordinate provided by docking and made this a parameter by Chemish. I improved docking score by this.

View full abstract

In this research, independent component analysis (ICA) and regression analysis are combined to extract significant components. ICA is a method that extracts mutually independent components from explanatory variables. We propose a new method that selects combination of independent components by using genetic algorithm (GA). It can construct a PLS model that has high predictive accuracy. This method is named ICA-GAPLS. In order to verify the superiority of ICA-GAPLS, this method is applied to QSPR analysis of aqueous solubility. The result of comparison with PLS and other regression methods is shown. For example, R2 and Q2 value in the first component of the PLS model are 0.427 and 0.421, respectively. These values in the ICA-GAPLS model are 0.945 and 0.857. ICA-GAPLS model achieves high predictive accuracy with less number of components than PLS model. ICA-GAPLS shows better result regarding the maximum of R2 and Q2 value than other methods. It is thought that ICA-GAPLS can extract effective components from explanatory variables and can construct the regression model having high predictive accuracy.

View full abstract