Transcript - ChemAxon
Leveraging ChemAxon Cheminformatics in an Integrated Drug Discovery and Development Platform Zhenbin Li, Paul Starbard, Jim Gregory, Donald Chen, Paul Bennett, Susan Virdee, Scott Sogge, and Charles Manly Neurogen Corporation, Branford, Connecticut, USA INTRODUCTION Neurogen is a small molecule drug discovery and development company targeting new drug candidates to improve the lives of patients suffering from pain, insomnia, inflammation, depression, and obesity. A reliable and high-performance compound library database, along with our sophisticated Accelerated Intelligent Drug Discovery (AIDDSM) system, is crucial for our compound screening, modification, and optimization to hit various biological targets, and ultimately enables us to deliver potent compounds into clinical trials. In 2003, Neurogen selected ChemAxon’s JChem as the backbone of our compound library. This poster presents some implementations of ChemAxon technology in our company. DESIGN AND IMPLEMENTATION Design Requirements 1. To centralize compound management that can store, search, and identify structure details with satisfying performance 2. All applications to register, track, search, display, and model compounds should be web-based 3. Compound information must be easy to leverage with existing Oracle-based research informatics system structural and sample attributes search criteria. This allows users to define any combination of search criteria. Structure-based searches are empowered by JChem. When structural information is cached in the system, structure-based search can be finished within seconds, if not in milliseconds. The application shown in Fig. 4 also provides options to display result information retrieved from system. Some of the options are from MarvinView. Fig. 2. Compound registration web interfaces Compound Data Browsing The compound management enables a better browsing environment for research data. As displayed in Fig. 5, structure searches yield a list of structures and sample lots, and each sample lot can directly link to assay data (i.e., dose response) in LIMS. The information becomes feedback for AIDDSM for further refine the models of compound design and optimization. Database Design Structure Cd_Id Cd_Structure … Structure_Id … … Sample_Structure Structure_Id Sample_id … … Sample Sample_id … … … … Lot Lot_id Sample_id … … … Salt_Structure Salt_id … … … Fig. 3. Rack and compound design web interface Fig. 1 Relational database tables of structures and samples A sample is a unique combination of structures, and lot is an instance of a sample. Salt form and crystal hydration status are tracked as attributes of lot. Structure and salt_structure are JChem tables containing detailed information about the structure as well as its fingerprints. Structure-related chemical properties are also stored in database. Oracle9i is the selected database engine, and it hosts all information about structures and other research data such as HTS, PK etc. Compound Searching Data retrieval from database is always a key to success. The usefulness of a sophisticated system is not only defined by how optimized the data have been stored, but, more importantly, is determined by how quickly and effectively to retrieve the information of interest. The web-based chemistry (Fig. 4) combines Data browser can also prioritize compounds for various screens, assay, or studies. Therefore compound management system is really the centralized system for managing, controlling and monitoring the processes and operations in drug discovery and development for the entire company. Compound Registration Our compound registration web interface (Fig. 2) allows scientists to design and register structures into our cheminformatics system. It uses both MarvinSketch and MarvinView. Add and Remove buttons allow the user to correct possible registration errors. The interface also allows registering more than one compound in a sample (mixtures). The interface on the right of Fig 2 is a summary page for the new lot. Structure and attributes will be inserted into the database when the user confirms the information. Not only can compounds be registered individually, a number of compounds can be registered as a group. Fig. 2 shows an interface to allow chemists to design structures based on chemistry rules. Therefore the compounds synthesized will potentially have better drug like properties. Fig. 5. Compound data browser interfaces SUMMARY ChemAxon’s cheminformatics provides a cost-effective and high-performance platform for compound management. Marvin applets and structure search tool are particularly useful, and are frequently employed in our applications. Based on our experiences with JChem, we would like to see more optimizations in areas such structure cleaning, and searching within the result set of a previous search or customized lists. Fig. 4. Compound search interface for defining search criteria Copyright @ 2005 Neurogen Corporation