bestClusterTreeThresholds {conos}

Find threshold of cluster detectability in trees of clusters

Description

For a given clustering, walks the walktrap (of clusters) result tree to find a subtree with max(min(sens,spec)) for each cluster, where sens is sensitivity, spec is specificity

Usage

bestClusterTreeThresholds(res, leaf.factor, clusters, clmerges = NULL)

Arguments

res

walktrap result object (igraph) where the nodes were clusters

leaf.factor

a named factor describing cell assignments to the leaf nodes (in the same order as res$names)

clusters

cluster factor

clmerges

integer matrix of cluster merges (default=NULL). If NULL, the function treeJaccard() performs calculation without it.

Value

a list of $thresholds - per cluster optimal detectability values, and $node - internal node id (merge row) where the optimum was found