Generic Chemical Structures Retrieval System Implemented By Dynamic-Link Library

The retrieval of pharmaceutical patent information plays an important role in the development of new drugs as well as protection of chemical patents. Generic structures, also called Markush structures, which are used widely in chemical patents, have always been one of the most challenging aspects of patent information retrieval due to their intrinsic complexity. The problem of how to represent generic structures by use of computers has been the focus of the filed of chemoinformatics. One practical retrieval system should be able to parse, code and store generic structures in such a way that it could avoid the problem of combinatorial explosion. The corresponding database underlying the retrieval system is much valuable for theoretical and practical research.In China, there is still no such a system for store and retrieval generic structures while more and more chemical patents have been granted, In this paper, the nature and theoretical foundations of generic structures devised during several research groups for the successful solution of the problem in order to provide a practical system are reviewed. A simple server-based retrieval system is designed, which accomplishes the following three functions, i.e., input, manipulation and output. A program, StruDraw which is implemented with Java applet accepts the input query structure. In this method, a chemical structure is regarded as an undirected graph with each atom being node and each bond being edge. Those nodes with connectivity degree less than two are recursively deleted thus resulting to a molecular scaffold which is a subgraph of the original chemical graph.The manipulation of structures are realized by Dynamic-Link Library(DLL), which has two interfaces with the front one to parse the output of the StruDraw. retrieve each structure stored in database and compare the two chemical graphs by use of back-tracking algorithm. The output of the final screening result is in the form of webpage.Compared with commonly used techniques, such as Common Gate Interface (CGI) or Active Server Pages (ASP), DLL is more secure, reliable and fast due to it is implemented with C++, making it more appropriate for building server-based applications. The designed system is tested on the platform of Windows XP with IIS5.0, which supports searches including specific and generic query structures. The results prove it suitable to screening of generic structures before the precise matching of atom-by-atom.