HierarchicalParam-class: The HierarchicalParam class

In LTLA/bluster: Clustering Algorithms for Bioconductor

The HierarchicalParam class

Description

The HierarchicalParam is a virtual subclass of the BlusterParam class. It causes clusterRows to dispatch to clustering algorithms that produce a dissimilarity matrix and a dendrogram.

Available slots

The virtual class provides metric, the choice of distance metric. This is conventionally passed to dist and defaults to a Euclidean distance in most subclasses. The distance function can be changed with the parameter dist.fun.

It also provides a number of slots to manage the final tree cut:

• cut.fun, a function that takes a hclust object as its first argument and returns a vector of cluster assignments. If NULL, the choice of function is determined from cut.dynamic.

• cut.dynamic, a logical scalar indicating whether a dynamic tree cut should be performed by cutreeDynamic. Otherwise cutree is used. Ignored if cut.fun is not NULL.

• cut.params, further arguments to pass to the tree cut function specified by the previous arguments.

Return value

The contract is that, when full=TRUE, the objects field of the clusterRows return value will always contain at least the following elements:

• dist, a dist object containing a dissimilarity matrix, usually a distance matrix.

• hclust, a hclust object containing a dendrogram.

See Also

HclustParam, for the archetypal example of a concrete subclass.

LTLA/bluster documentation built on Sept. 8, 2024, 4:37 a.m.