Automated search for potentially active compounds by using cluster trees

摘要: Abstract A new agglomerative (bottom-up) hierarchical cluster technique referred to as the Adaptive Mean-Linkage algorithm is derived. Cluster algorithms are also offered a tool explore descriptor space knowing quantitative structure–activity relationship (QSAR). The substituents clustered building dendrogram (cluster tree) per site. Choosing appropriate pathways on such trees according QSAR equation, an automated search for potentially active substituted compounds can be performed. Applications series of phenylguanidines with anticancer activity focused illustrating this approach.