find_clones: Identify nodes in a hierarchical tree which qualify as...

In seqpipe/SCclust: Clustering of Single Cell Sequencing Copy Number Profiles to Identify Clones

Description

Identify nodes in a hierarchical tree which qualify as clones. Identify 'hard' clones first, then expand them to 'soft' clones. Expansion may result in clone mergers. Based on hierarchical clustering, identify the hard/soft clones.

Usage

find_clones(hc, fdrthresh = -2, sharemin = 0.85, nshare = 3, bymax = T,
  climbfromsize = 2, climbtoshare = 3)

Arguments

hc	An hclust object with additional items generated by hclust_tree.
fdrthresh	maximal allowed value for log10(FDR) for any pair of leaves in a clone node. Default: -2.
sharemin	A feature is considered 'widely shared' if present in sharemin fraction of leaves in a node.Default: 0.90.
nshare	Minimal number of 'widely shared' features in a hard clone. Default: 3.
bymax	Logical. If TRUE (default), use maximal, and otherwise mean, FDR for the node as a criterion for a hard clone.
climbfromsize	An integer: minimal size of a hard clone allowed to be expanded
climbtoshare	An integer: expand the clone as long as the number of widely shared features is at least this value

Value

An hclust object, with hard/soft clones indicated

seqpipe/SCclust documentation built on May 4, 2021, 11:56 a.m.