rubikclust: Unsupervised clustering to detect rough structures and...

In kridsadakorn/kris: Keen and Reliable Interface Subroutines for Bioinformatic Analysis

Unsupervised clustering to detect rough structures and outliers.

Description

Handle and operate on Nx3 matrix, where N is the number of samples and data are collected on 3 variables.

Usage

rubikclust(X, min.space = 0.4, rotation = TRUE)

Arguments

X	A data matrix for which rows represent samples and the 3 columns represent features. Missingness is not allowed.
min.space	A value to specify a minimum space between 2 consecutive projected values. Default = 0.4.
rotation	To specify if rotation is enabled or not. Default = TRUE.

Details

The function rubikClust is able to take up to 3 variables (N x 3 matrix). In case, a matrix contains more than 3 columns, only the first three columns are used; the other columns are ignored.

Value

The returned value is a vector of numbers representing cluster memberships.

Examples

#Load simulated dataset
data(example_SNP)

PCs <- cal.pc.linear(simsnp$snp, no.pc = 3)

#Run rubikclust with the default parameters
groups <- rubikclust(PCs$PC)
#Check clustering results
print(groups)

#Check cluster's distribution
table(groups)

#Check the plot, highlight the points according to the clustering result
mylabels <- paste0("group", as.factor(groups))
plot3views( PCs$PC, labels = mylabels)

#Run rubikclust with min.space = 0.02
groups <- rubikclust(PCs$PC, min.space = 0.02)
#Check clustering results
print(groups)

#Check cluster's distribution
table(groups)

#Check the plot, highlight the points according to the clustering result
mylabels <- paste0("group", as.factor(groups))
plot3views( PCs$PC, labels = mylabels)

kridsadakorn/kris documentation built on April 28, 2022, 4:23 a.m.