biclustFunc: biclustFunc function

In ipADMIXTURE: Iterative Pruning Population Admixture Inference Framework

Description

biclustFunc is a binary clustering function using hierarchical clustering.

Usage

biclustFunc(Qmat, admixRatioThs = 0.5, method = "average")

Arguments

Qmat	is a Q matrix that contains admixture ratios of all individuals where the Qmat[i,j] represents the admixture ratio of ancestor j for individual i.
admixRatioThs	is a threshold to determine that if a cluster has maxDiffAdmixRatio lower than threshold, then the cluster is a homogeneous cluster.
method	is a method parameter of hclust object for hierarchical clustering analysis. The default is "average".

Value

This function returns binary clustering results.

heteroFlag	is a flag that represents a status whether a given cluster is heterogeneous (having sub-clusters). It is TRUE if maxDiffAdmixRatio >= admixRatioThs.
clusterInx	is a vector of clustering assignment where indexClsVec[i] is a cluster number of individual i.
meanDiffAdmixRatio	is a vector of magnitude-difference of admixture ratios. It is calculated by splitting a given cluster into two sub-clusters. Then, we take the absolute on the difference between mean admixture ratios of sub-clusters.
Qmat1	is a Q matrix of sub-cluster #1 after splitting a given cluster into two sub-clusters that contains admixture ratios of all individuals where the Qmat[i,j] represents the admixture ratio of ancestor j for individual i.
Qmat2	is a Q matrix of sub-cluster #2 after splitting a given cluster into two sub-clusters that contains admixture ratios of all individuals where the Qmat[i,j] represents the admixture ratio of ancestor j for individual i.
maxDiffAdmixRatio	is a maximum of magnitude-difference of admixture ratios for a given cluster before splitting into two sub-clusters.

Examples

# Running biclustFunc on Q matrix of 27 human population dataset where K = 12
obj<-biclustFunc(Qmat=ipADMIXTURE::human27pop_Qmat[[11]], admixRatioThs =0.15)

ipADMIXTURE documentation built on March 26, 2020, 7:38 p.m.