Substructural Analysis of Antineoplastic Compounds in Respect of in Vivo Tumor Models
Ilza K. Pajeva, Evgeny V. Golovinsky · Quantitative Structure-Activity Relationships · 1990
Abstract A substructural analysis of antineoplastic compounds is provided with respect to two in vivo tumor models: lymphoid leukemia L1210 (L1210) and Lewis lung carcinoma (LL). The activity data are expressed by 4 levels of activity against L1210 and LL. The structures of the compounds are described by means of structural features derived from a structure‐activity relationship (SAR) oriented language for two‐dimensional chemical structure representation. A statistical‐heuristic technique is applied to obtain, for each structural feature, a weight for each activity level. On the basis of the weights, the contributions of the structural features to the probabilities of activity against L1210 and LL models are estimated. The features contributing to the relevant probabilities to the greatest extent are located and analyzed from a biochemical point of view. The results obtained show the suitability of the features for compounds' activity recognition and identification of substructures significant for the investigated activities.