nlnet (version 1.4)

nlhc: Non-Linear Hierarchical Clustering

Description

The non-linear hierarchical clustering based on DCOL

Usage

nlhc(array, hamil.method = "nn", concorde.path = NA, 
use.normal.approx = FALSE, normalization = "standardize", combine.linear = TRUE,
use.traditional.hclust = FALSE, method.traditional.hclust = "average")

Arguments

array

the data matrix with no missing values

hamil.method

the method to find the hamiltonian path.

concorde.path

If using the Concorde TSP Solver, the local directory of the solver

use.normal.approx

whether to use the normal approximation for the null hypothesis.

normalization

the normalization method for the array.

combine.linear

whether linear association should be found by correlation to combine with nonlinear association found by DCOL.

use.traditional.hclust

whether traditional agglomerative clustering should be used.

method.traditional.hclust

the method to pass on to hclust() if traditional method is chosen.

Value

Returns a hclust object same as the output of hclust(). Reference: help(hclust)

merge

an n-1 by 2 matrix. Row i of merge describes the merging of clusters at step i of the clustering. If an element j in the row is negative, then observation -j was merged at this stage. If j is positive then the merge was with the cluster formed at the (earlier) stage j of the algorithm.

height

a set of n-1 real values, the value of the criterion associated with the clusterig method for the particular agglomeration

order

a vector giving the permutation of the original observations suitable for plotting, in the sense that a cluster plot using this ordering and matrix merge will not have crossings of the branches.

labels

labels for each of the objects being clustered

call

the call which produced the result

dist.method

the distance that has been used to create d

height.0

original calculation of merging height

Details

Hamil.method: It is passed onto the function tsp of library TSP. To use linkern method, the user needs to install concord as instructed in TSP.

use.normal.approx: If TRUE, normal approximation is used for every feature, AND all covariances are assumed to be zero. If FALSE, generates permutation based null distribution - mean vector and a variance-covariance matrix.

normalization: There are three choices - "standardize" means removing the mean of each row and make the standard deviation one; "normal_score" means normal score transformation; "none" means do nothing. In that case we still assume some normalization has been done by the user such that each row has approximately mean 0 and sd 1.

combine.linear: The two pieces of information is combined at the start to initiate the distance matrix.

References

http://www.ncbi.nlm.nih.gov/pubmed/24334400

See Also

data.gen

Examples

# NOT RUN {
 ## generating the data matrix & hiden clusters as a sample
 input<-data.gen(n.genes=40, n.grps=4)
 ## now input includes data matrix and hiden clusters, so get the matrix as input.
 input<-input$data

 nlhc.data<-nlhc(input)
 plot(nlhc.data)
 ##get the merge from the input.
 merge<-nlhc.data$merge 
# }