WGCNA (version 1.43)

consensusProjectiveKMeans: Consensus projective K-means (pre-)clustering of expression data

Description

Implementation of a consensus variant of K-means clustering for expression data across multiple data sets.

Usage

consensusProjectiveKMeans(
  multiExpr, 
  preferredSize = 5000, 
  nCenters = NULL, 
  sizePenaltyPower = 4,
  networkType = "unsigned", 
  randomSeed = 54321,
  checkData = TRUE,
  useMean = (length(multiExpr) > 3),
  maxIterations = 1000, 
  verbose = 0, indent = 0)

Arguments

multiExpr

expression data in the multi-set format (see checkSets). A vector of lists, one per set. Each set must contain a component data that contains the expression data, with rows corresponding to s

preferredSize

preferred maximum size of clusters.

nCenters

number of initial clusters. Empirical evidence suggests that more centers will give a better preclustering; the default is as.integer(min(nGenes/20, preferredSize^2/nGenes)) and is an attempt to arrive at a reasonable number given the resou

sizePenaltyPower

parameter specifying how severe is the penalty for clusters that exceed preferredSize.

networkType

network type. Allowed values are (unique abbreviations of) "unsigned", "signed", "signed hybrid". See adjacency.

randomSeed

integer to be used as seed for the random number generator before the function starts. If a current seed exists, it is saved and restored upon exit.

checkData

logical: should data be checked for genes with zero variance and genes and samples with excessive numbers of missing samples? Bad samples are ignored; returned cluster assignment for bad genes will be NA.

useMean

logical: should mean distance across sets be used instead of maximum? See details.

maxIterations

maximum iterations to be attempted.

verbose

integer level of verbosity. Zero means silent, higher values make the output progressively more and more verbose.

indent

indentation for diagnostic messages. Zero means no indentation, each unit adds two spaces.

Value

• A list with the following components:
• clusters
  a numerical vector with one component per input gene, giving the cluster number in which the gene is assigned.
• centers
  a vector of lists, one list per set. Each list contains a component data that contains a matrix whose columns are the cluster centers in the corresponding set.
• unmergedClusters
  a numerical vector with one component per input gene, giving the cluster number in which the gene was assigned before the final merging step.
• unmergedCenters
  a vector of lists, one list per set. Each list contains a component data that contains a matrix whose columns are the cluster centers before merging in the corresponding set.

Details

The principal aim of this function within WGCNA is to pre-cluster a large number of genes into smaller blocks that can be handled using standard WGCNA techniques.

This function implements a variant of K-means clustering that is suitable for co-expression analysis. Cluster centers are defined by the first principal component, and distances by correlation. Consensus distance across several sets is defined as the maximum of the corresponding distances in individual sets; however, if useMean is set, the mean distance will be used instead of the maximum. The distance between a gene and a center of a cluster is multiplied by a factor of $max(clusterSize/preferredSize, 1)^{sizePenaltyPower}$, thus penalizing clusters whose size exceeds preferredSize. The function starts with randomly generated cluster assignment (hence the need to set the random seed for repeatability) and executes interations of calculating new centers and reassigning genes to nearest (in the consensus sense) center until the clustering becomes stable. Before returning, nearby clusters are iteratively combined if their combined size is below preferredSize.

Consensus distance defined as maximum of distances in all sets is consistent with the approach taken in blockwiseConsensusModules, but the procedure may not converge. Hence it is advisable to use the mean as consensus in cases where there are multiple data sets (4 or more, say) and/or if the input data sets are very different.

The standard principal component calculation via the function svd fails from time to time (likely a convergence problem of the underlying lapack functions). Such errors are trapped and the principal component is approximated by a weighted average of expression profiles in the cluster. If verbose is set above 2, an informational message is printed whenever this approximation is used.

See Also

projectiveKMeans