liveBook · Manning

Ward's method

As explained in Build AI Drug Discovery Pipelines

Ward's method is an agglomerative linkage criterion that merges the pair of clusters producing the smallest increase in within-cluster variance.

Ward's method is an agglomerative linkage criterion for grouping compounds. Starting with one compound per cluster, it repeatedly merges the pair that causes the smallest increase in total within-cluster variance. section 7.2.3

It is popular in cheminformatics, but it minimizes Euclidean variance, which may poorly fit sparse binary fingerprints. Tanimoto-based Taylor-Butina clustering can provide a more chemically appropriate comparison. section 7.2.3

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains variance-based merging and demonstrates fitting a hierarchy, while deferring final cluster selection to cutting the dendrogram.
Examples
Uses RDKit fingerprints, a dendrogram, sixteen-cluster sampling, and diversity or redundancy comparisons with Taylor-Butina clustering.
What you can do after
Apply Ward clustering with RDKit and scikit-learn to organize compounds and sample representatives from defined clusters.
How it connects
Connects agglomerative clustering, hierarchical clustering, Euclidean variance, molecular fingerprints, dendrograms, and Taylor-Butina clustering.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.