Premium
Structure‐activity relationship approaches and applications
Author(s) -
Tong Weida,
Welsh William J.,
Shi Leming,
Fang Hong,
Perkins Roger
Publication year - 2003
Publication title -
environmental toxicology and chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.1
H-Index - 171
eISSN - 1552-8618
pISSN - 0730-7268
DOI - 10.1897/01-198
Subject(s) - quantitative structure–activity relationship , computer science , ordinal scale , representation (politics) , scale (ratio) , machine learning , artificial intelligence , biochemical engineering , mathematics , engineering , statistics , physics , quantum mechanics , politics , political science , law
Abstract New techniques and software have enabled ubiquitous use of structure‐activity relationships (SARs) in the pharmaceutical industry and toxicological sciences. We review the status of SAR technology by using examples to underscore the advances as well as the unique technical challenges. Applying SAR involves two steps: Characterization of the chemicals under investigation, and application of chemometric approaches to explore data patterns or to establish the relationships between structure and activity. We describe generally but not exhaustively the SAR methodologies popular use in toxicology, including representation of chemical structure, and chemometric techniques where models are both unsupervised and supervised. The utility of SAR technology is most evident when supervised methods are used to predict toxicity of untested chemicals based only on chemical structure. Such models can predict on both an ordinal scale (e.g., active vs inactive) or a continuous scale (e.g., median lethal dose [LD50] dose). The reader is also referred to a companion paper in this issue that discusses quantitative structure‐activity relationship (QSAR) methods that have advanced markedly over the past decade.