The Synthesis Design System WODCA as a Potential Front-End to Daylight Reaction Databases
Molecular Networks GmbH, Computerchemie, Nägelsbachstraße 25 D-91052 Erlangen
A new release of the synthesis design system WODCA (Workbench for the Organization of Data for Chemical Applications) developed during the last eleven years in the workgroup of Prof. Gasteiger at the Computer-Chemie-Centrum in Erlangen is to be presented.
WODCA V.5 implements several new approaches to the retrosynthetic analysis of target compounds which are derived by means of a link to reaction databases. In the current implementation the reaction databases have to be transformed into WODCA’s own reaction database format. If it is of interest to the Daylight user community a direct connection could also be established to interface WODCA to Daylight reaction databases.
Two different approaches have been developed in the current implementation.
In the first approach the disconnection of a target compound is performed by WODCA with one of its disconnection strategies. The result of such a disconnection (product Þ reactants) is used to derive a query for a reaction substructure search. The benefit is that the users do not have to define the queries on their own, which is often an error-prone and time consuming task. As a result the disconnection of the target is verified by a reaction database. All information which is stored in the reaction database (yields, reaction conditions, literature) can be displayed in WODCA’s reaction browser.
In the second approach the target compound is analysed by means of neural networks. For this purpose WODCA has been combined with the CORA (Classification of Organic Reactions for Applications) system. Within CORA the target compound is compared to the products of all reactions from a reaction database which are classified and stored in a neural network. CORA analyses the target compound for strategic bonds which have similar physicochemical properties to the products of reactions in the reaction database. Thus, the search is performed in a neural network in contrast to the classical structure search. The benefit of this approach is the consideration of reaction types not included in the existing disconnection strategies of WODCA (e.g. multistep reactions, pericyclic reactions, rearrangements, catalytic reactions). As a result, all the information which is stored in the neural network (yields, reaction conditions, literature) can be displayed in WODCA’s reaction browser by retrieval of the record numbers of close reactions as classified by the neural network.
Some of the new functionalities of WODCA will be presented in an on-line demonstration.
Daylight Chemical Information
Systems, Inc.
info@daylight.com