Vol.2 No.1 2009

67/88

Research paper−64−Synthesiology - English edition Vol.2 No.1 pp.64-72 (Jun. 2009) based on several million compounds marketed every year and to make it immediately available to the pharmaceutical industry. Several dozen major reagent vendors in the world distribute electronic files of reagent catalogs listing their 2D molecular structures, but 3D stereoscopic molecular structures rather than 2D chemical structures are necessary for VS. Therefore, the purpose of our study was to construct 3D molecular structures from the 2D molecular structures of millions of compounds in these catalogs and distribute them by compiling a database.3 BenefitsIrrespective of the method used, the development of pharmaceutical compounds begins with a search for candidate compounds that could bind to the target proteins from compound databases. Computational searching in compound databases is a necessity in the modern drug discovery process. However, there are several issues in this process:(1) Although compound DBs for VS have been developed and marketed for pharmaceutical companies by overseas software developers since the 1980s, license fees are expensive, with a license costing 4–6 million yen per year[1].(2) Software products for the development of compound DBs are also marketed by overseas software developers[2]. Having used costly software to develop compound DBs for VS, we experienced several issues in terms of their quality such as frequent representations of incorrect 3D molecular structures, incorrect appositions of hydrogen atoms, and the generation 1 IntroductionOne of the primary objectives in the post-genome era is the innovation of drug discovery. However, compared with the dramatic advancements in genetic analysis technology, drug discovery processes have been experiencing continuous difficulties and the expected results have not been achieved. In this situation, computational drug screening (in-silico or virtual screening (VS)) is considered to be one of the strategies for streamlining the drug discovery process. VS is used to computationally select seed molecules from existing molecules for pharmaceutical applications. Thus, VS requires a computationally accessible database of chemical compounds with 3D molecular structures (hereafter referred to as “compound DB”). Although overseas compound DB products are available, we developed an in-house DB due to the issues of price, quality, and management of results. We developed a compound DB by eliminating duplicated data using a chemical informatics approach, constructing 3D compound structures by a molecular force-field method, and computing atomic charges by quantum chemical calculations. These methods are described in chapter 4. In addition, we also developed a novel DB that predicts the binding energy of large numbers of predefined proteins and compounds. Using these DBs, it is now possible to predict active compounds with respect to target proteins for drug discovery with a high probability of success.2 ObjectiveOur objective was to develop a compound DB usable for VS - Development of 3D-compound database -Yoshifumi Fukunishi1*, Yuusuke Sugihara2, Yoshiaki Mikami3, Kohta Sakai2, Hiroshi Kusudo3 and Haruki Nakamura41. Biomedicinal Information Research Center, AIST Aomi 2-41-6, Koto-ku 135-0064, Japan *E-mail : y-fukunishi@aist.go.jp, 2. FUJITSU KYUSHU SYSTEM ENGINEERING LIMITED Life Science System Dept. PLM Solution Div. Nakase 1-9-3, Mihama-ku, Chiba 261-8588, Japan, 3. Hitachi East Japan Solutions, Ltd. Ekimaehonchou 12-1, Kawasaki-ku, Kawasaki 210-0007, Japan, 4. Institute for Protein Research, Osaka University Yamadaoka 3-2, Suita 565-0871, JapanReceived original manuscript October 28,2008, Revisions received December 9,2008, Accepted December 9,2008Every year, several million compounds for drug screening are released by numerous vendors around the world. The information provided on these compounds is in the form of their two-dimensional (2D) structures. We have developed a software system to generate a database of three-dimensional (3D) structures of these compounds and have distributed this database. We have also developed a database of protein-compound docking scores of 180 proteins with respect to these millions of compounds. These databases make it possible to identify new active compounds for many drug targets.Advanced in-silico drug screening to achieve high hit ratioKeywords : compound database, myPresto, virtual screening, in-silico drug screening, compound library[Translation from Synthesiology, Vol.2, No.1, p.60-68 (2009)]

※このページを正しく表示するにはFlashPlayer9以上が必要です