Article ID: 210 - Last Modified: April 30, 2012
Does Schrödinger software include scaffold clustering ?
There is not currently such a feature. However, Canvas, the cheminformatics platform, has some similarly useful tools.
First, try CanvasMCS, the maximum common substructure facility. You can specify a minimum and maximum number of structures that must match in order to find common scaffolds in a dataset. Use the checkbox "Enforce minimum size" to avoid generating small substructures.
For up to a few thousand structures, you can also try Hierarchical Clustering. After the initial clustering step, select clusters at the Kelley level. Every cluster may not share a common scaffold, but this method does a very good job of separating compounds into different chemical classes.
For larger numbers of compounds, there is Leader-Follower clustering. Experiment with the size of the cluster radius to get clusters that are sufficiently homogeneous without a large population of singletons. There is also K-Means clustering. You can specify the number of clusters in advance.
Finally, the R-Group Analysis tool can also be useful. In Suite 2010. this tool is included as a Workflow. From Suite 2011 it is available in both Maestro (Tools menu) and Canvas (Scripts menu).
Keywords: Canvas, clustering
Type the words or phrases on which you would like to search, or click here to view a list of all
Knowledge Base articles