作者: Souâd Mghazli , Abderrahim Jaouad , Mohamed Mansour , Didier Villemin , Driss Cherqaoui
DOI: 10.1016/S0045-6535(00)00111-9
关键词:
摘要: Abstract Models of relationships between structure and antifungal activity 1-[2-(substituted phenyl)allyl]imidazoles related compounds were constructed by means a multilayer neural network using the back-propagation (BP) algorithm. Each molecule was described three structural one physicochemical parameters. The leave-one-out procedure used to assess predictive ability model. results obtained compared those given in literature multiple linear regression (MLR), found be better. contribution each descriptor structure–activity evaluated. Hydrophobicity confirmed take most relevant part molecular description.