Learn More Web Services Manual Clustering Manual

Scaffold-Directed Clustering Clusters characterized by a common substructure The new scaffold-directed clustering is available both as a Daylight Web Service and as part of the Daylight Clustering Package. It is distinct from many other clustering methods in that it generates clusters of compounds, where each cluster is represented by a scaffold. The scaffold is a SMARTS® expression with one or more fragments that contains the structural features common to all members of the cluster. Clustering is directed in such a way as to guarantee a defined level of scaffold coverage for all members of a cluster. It is an ideal clustering method for applications involving smaller data sets such as determining structure-activity relationships within lead optimization projects or analysis of structures provided in publications or patents. The method differs from tree-based scaffold analysis methods in that 1) it does not use heuristics to generate scaffolds, 2) it does not use any predefined fragments, and 3) it generates a single set of clusters that meet an intuitive coverage metric. The scaffold-directed method is therefore more general and directly provides sets of compounds ready for further analysis.