Publication Details (including relevant citation information):
J. Chem. Inf. Comput. Sci., 1993, 33 (6), pp. 886-895.
A novel knowledge-based data evaluation system for organic compounds, DESOC, has been developed. DESOC comprises a compound/property database and several modules for the prediction of properties. The database contains compounds that are represented by SMILES notations. Property data are Antoine coefficients for temperature-dependent vapor pressure calculations, aqueous solubility data at 20 and 25 ºC, and partition coefficients for the system air/water, 1-octanol/water, and soil/water. DESOC includes routines for data retrieval, similarity-based search, and property estimation. Estimation of a query property is based on (1) identification of database compounds structurally related to the query, (2) recognition of the structural difference between query and database compounds, and (3) translation of the structural difference into the corresponding property difference. A new approach, the group interchange method (GIM), is introduced for the representation and analysis of structural differences between similar compounds. Compounds are related to each other in terms of elementary group operations. These operations are encoded as linear notations using the grammar of SMILES. The performance of DESOC is illustrated by protocol files generated for selected queries.
Address (URL): http://dx.doi.org/10.1021/ci00016a011