Abstract
Fifteen quantitative structure-activity relationship (QSAR) models developed by various authors for the prediction of mutagenicity of aromatic and heteroaromatic amines were analyzed and thirteen of them, based on 95 amines, were compared using their respective statistics and order theory (Hasse Diagram Technique, HDT) to obtain an ordering of QSAR models. The technique of Formal Concept Analysis (FCA) was applied to the set of 95 amines to extract concepts and, in general, knowledge about the relationship between structural attributes and mutagenicity. HDT may be useful as a general tool for the comparison of different classes of QSAR models. FCA turns out to be a novel mathematical technique for seeking for relationships between molecular structure and activity.
Keywords: Mutagenicity, quantitative structure-activity relationship (QSAR), structure-activity relationship (SAR), Hasse diagram technique (HDT), formal concept analysis (FCA), aromatic amines, molecular descriptors, neural networks, heteroaromatic amines, QSAR models
9