COMPUTER-ASSISTED STUDIES OF CHEMICAL STRUCTURE - BIOLOGICAL ACTIVITY RELATIONSHIPS (PATTERN RECOGNITION, ANTIINFLAMMATORY, QSAR, CHEMOMETRICS, COMPUTATIONAL)

This work is divided into four parts. First is a pattern recognition (PR) investigation of the relationship between chemical structure and chemical mutagenesis. This was an effort to develop an automated screening system for the detection of potential mutagens. Second is an in-depth investigation of the reliability of the results provided by PR methods in order to define the limitations of these techniques and better understand how to apply them to structure-activity relationship (SAR) problems. Third are SAR studies of anti-inflammatory steroids in an effort to understand the structural features that control their anti-inflammatory activity. Fourth is a description of a computerized method for the definition, manipulation, and comparison of molecular shape, an important property which affects a compound's other chemical and biological properties. With this method, shape differences can be rigorously incorporated into PR-SAR studies.