Premium
Nonempirical approach to structure–activity studies
Author(s) -
Randić Milan
Publication year - 1984
Publication title -
international journal of quantum chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.484
H-Index - 105
eISSN - 1097-461X
pISSN - 0020-7608
DOI - 10.1002/qua.560260716
Subject(s) - weighting , analogy , computer science , set (abstract data type) , cluster analysis , graph , path (computing) , parametrization (atmospheric modeling) , theoretical computer science , mathematics , computational chemistry , artificial intelligence , chemistry , physics , quantum mechanics , epistemology , philosophy , acoustics , programming language , radiative transfer
After a brief review of empirical and nonempirical schemes used in the study of molecular properties, the difficulties associated with structure–activity correlations are outlined. Part of the difficulty originates with the lack of precise definition of a structure. An analogy is made with similarly vague concepts of structural chemistry—the notion of aromaticity. This follows with the description of a characterization of structures by selected graph invariants. In particular we consider the count of weighted paths derived from suitable weighting of the individual bonds in a structure. We adopted the weighting factors ( mn ) −1/2 introduced originally for the definition of the connectivity index. The approach thus combines some features of the very successful connectivity index with features of path sequences, found very useful in comparative studies of related compounds. As an illustration of the combined approach we consider a set of some 40 therapeutically active substances, studied previously by others, and derive their clustering (classification) which is solely based on the count of weighted paths and is devoid of any empirical parametrization.