During the last two decades many graph invariants have been developed and used for predicting properties of molecules. However, no systematic study has been undertaken to test the effectiveness of these parameters in correlating properties of large and diverse groups of molecules. We have used a large set of graph invariants in the prediction of lipophilicity (log P, octanol-water) of a number of large and diverse databases of chemicals. The utility of graph theoretic indices in structure-activity relationship (SAR) and property-activity relationship (PAR) in general is discussed. Correlations of lipophilicity with calculated graph invariants is analyzed in terms of molecular factors that determine this property.
Bibliographical noteFunding Information:
The authors wish to express their appreciation to the U.S. Environmental Protection Agency (Contract CR-815387) for support. Thanks to Anda Bellamy and Greg Grunwald for their assistance in the project. This paper is contribution nuonber 52 of the Center for Water and the Environment.
- Graph invariants
- connectivity indices
- information-theoretic indices
- structure-activity relationship (SAR)
- topological indices