Computer-assisted studies of molecular structure and carcinogenic activity.

Both quantitative and qualitative studies of the structure-activity relationships (SAR) of carcinogenic compounds are discussed. Using the ADAPT software system, examples are given of (1) predicting the gas and liquid chromatographic retention behavior of polycyclic aromatic compounds, (2) predicting the carcinogenicity of N-nitrosamines, and (3) predicting the mutagenic behavior of pesticides and other compounds in short-term test systems. It is concluded that the complexity and lack of understanding of the carcinogenic response should not prevent scientists from searching for useful empirical relationships between molecular structure and carcinogenicity. Computer generated molecular descriptors and statistical and pattern recognition techniques can be useful tools in this research.