ClusterExperiment-class: Class ClusterExperiment

In epurdom/clusterExperiment: Compare Clusterings for Single-Cell Sequencing

Class ClusterExperiment

Description

ClusterExperiment is a class that extends SingleCellExperiment and is used to store the data and clustering information.

In addition to the slots of the SingleCellExperiment class, the ClusterExperiment object has the additional slots described in the Slots section.

There are several methods implemented for this class. The most important methods (e.g., clusterMany, makeConsensus, ...) have their own help page. Simple helper methods are described in the Methods section. For a comprehensive list of methods specific to this class see the Reference Manual.

The constructor ClusterExperiment creates an object of the class ClusterExperiment. However, the typical way of creating these objects is the result of a call to clusterMany or clusterSingle.

Note that when subsetting the data, the co-clustering and dendrogram information are lost.

Usage

ClusterExperiment(object, clusters, ...)

## S4 method for signature 'matrixOrHDF5,ANY'
ClusterExperiment(object, clusters, ...)

## S4 method for signature 'SummarizedExperiment,ANY'
ClusterExperiment(object, clusters, ...)

## S4 method for signature 'SingleCellExperiment,numeric'
ClusterExperiment(object, clusters, ...)

## S4 method for signature 'SingleCellExperiment,character'
ClusterExperiment(object, clusters, ...)

## S4 method for signature 'SingleCellExperiment,factor'
ClusterExperiment(object, clusters, ...)

## S4 method for signature 'SingleCellExperiment,matrix'
ClusterExperiment(
  object,
  clusters,
  transformation = function(x) {     x },
  primaryIndex = 1,
  clusterTypes = "User",
  clusterInfo = NULL,
  orderSamples = seq_len(ncol(object)),
  dendro_samples = NULL,
  dendro_index = NA_real_,
  dendro_clusters = NULL,
  coClustering = NULL,
  merge_index = NA_real_,
  merge_cutoff = NA_real_,
  merge_dendrocluster_index = NA_real_,
  merge_nodeProp = NULL,
  merge_nodeMerge = NULL,
  merge_method = NA_character_,
  merge_demethod = NA_character_,
  clusterLegend = NULL,
  checkTransformAndAssay = TRUE
)

Arguments

object	a matrix or SummarizedExperiment or SingleCellExperiment containing the data that was clustered.
clusters	can be either a numeric or character vector, a factor, or a numeric matrix, containing the cluster labels.
...	The arguments transformation, clusterTypes and clusterInfo to be passed to the constructor for signature SingleCellExperiment,matrix.
transformation	function. A function to transform the data before performing steps that assume normal-like data (i.e. constant variance), such as the log.
primaryIndex	integer. Sets the 'primaryIndex' slot (see Slots).
clusterTypes	a string describing the nature of the clustering. The values 'clusterSingle', 'clusterMany', 'mergeClusters', 'makeConsensus' are reserved for the clustering coming from the package workflow and should not be used when creating a new object with the constructor.
clusterInfo	a list with information on the clustering (see Slots).
orderSamples	a vector of integers. Sets the 'orderSamples' slot (see Slots).
dendro_samples	phylo4 object. Sets the 'dendro_samples' slot (see Slots).
dendro_index	numeric. Sets the dendro_index slot (see Slots).
dendro_clusters	phylo4 object. Sets the 'dendro_clusters' slot (see Slots).
coClustering	matrix. Sets the coClustering slot (see Slots).
merge_index	integer. Sets the merge_index slot (see Slots)
merge_cutoff	numeric. Sets the merge_cutoff slot (see Slots)
merge_dendrocluster_index	integer. Sets the merge_dendrocluster_index slot (see Slots)
merge_nodeProp	data.frame. Sets the merge_nodeProp slot (see Slots)
merge_nodeMerge	data.frame. Sets the merge_nodeMerge slot (see Slots)
merge_method	character, Sets the merge_method slot (see Slots)
merge_demethod	character, Sets the merge_demethod slot (see Slots)
clusterLegend	list, Sets the clusterLegend slot (see details).
checkTransformAndAssay	logical. Whether to check the content of the assay and given transformation function for whether they are valid.

Details

The clusterLegend argument to ClusterExperiment must be a valid clusterLegend format and match the values in clusters, in that the "clusterIds" column must matches the value in the clustering matrix clusters. If names(clusterLegend)==NULL, it is assumed that the entries of clusterLegend are in the same order as the columns of clusters. Generally, this is not a good way for users to set the clusterLegend slot.

The ClusterExperiment constructor function gives clusterLabels based on the column names of the input matrix/SingleCellExperiment. If missing, will assign labels "cluster1","cluster2", etc.

Note that the validity check when creating a new ClusterExperiment object with new is less extensive than when using ClusterExperiment function with checkTransformAndAssay=TRUE (the default). Users are advised to use ClusterExperiment to create new ClusterExperiment objects.

Value

A ClusterExperiment object.

Slots

transformation

function. Function to transform the data by when methods that assume normal-like data (e.g. log)

clusterMatrix

matrix. A matrix giving the integer-valued cluster ids for each sample. The rows of the matrix correspond to clusterings and columns to samples. The integer values are assigned in the order that the clusters were found, if found by setting sequential=TRUE in clusterSingle. "-1" indicates the sample was not clustered.

primaryIndex

numeric. An index that specifies the primary set of labels.

clusterInfo

list. A list with info about the clustering. If created from clusterSingle, clusterInfo will include the parameter used for the call, and the call itself. If sequential = TRUE it will also include the following components.

• clusterInfoif sequential=TRUE and clusters were successfully found, a matrix of information regarding the algorithm behavior for each cluster (the starting and stopping K for each cluster, and the number of iterations for each cluster).

• whyStopif sequential=TRUE and clusters were successfully found, a character string explaining what triggered the algorithm to stop.

merge_index

index of the current merged cluster

merge_cutoff

value for the cutoff used to determine whether to merge clusters

merge_dendrocluster_index

index of the cluster merged with the current merge

merge_nodeMerge

data.frame of information about nodes merged in the current merge. See mergeClusters

merge_nodeProp

data.frame of information of proportion estimated non-null at each node of dendrogram. See mergeClusters

merge_method

character indicating method used for merging. See mergeClusters

merge_demethod

character indicating the DE method used for merging. See mergeClusters

clusterTypes

character vector with the origin of each column of clusterMatrix.

dendro_samples

phylo4d object. A dendrogram containing the cluster relationship (leaves are samples; see clusterDendrogram for details).

dendro_clusters

phylo4d object. A dendrogram containing the cluster relationship (leaves are clusters; see see sampleDendrogram for details).

dendro_index

numeric. An integer giving the cluster that was used to make the dendrograms. NA_real_ value if no dendrograms are saved.

coClustering

One of

• NULL, i.e. empty

• a numeric vector, signifying the indices of the clusterings in the clusterMatrix that were used for makeConsensus. This allows for the recreation of the distance matrix (using hamming distance) if needed for function plotClusters but doesn't require storage of full NxN matrix.

• a sparseMatrix object – a sparse representation of the NxN matrix with the cluster co-occurrence information; this can either be based on subsampling or on co-clustering across parameter sets (see clusterMany). The matrix is a square matrix with number of rows/columns equal to the number of samples.

clusterLegend

a list, one per cluster in clusterMatrix. Each element of the list is a matrix with nrows equal to the number of different clusters in the clustering, and consisting of at least two columns with the following column names: "clusterId" and "color".

orderSamples

a numeric vector (of integers) defining the order of samples to be used for plotting of samples. Usually set internally by other functions.

See Also

sparseMatrix phylo4d

Examples

sce <- matrix(data=rnorm(200), ncol=10)
labels <- gl(5, 2)

cc <- ClusterExperiment(sce, as.numeric(labels), transformation =
function(x){x})

epurdom/clusterExperiment documentation built on April 23, 2024, 9:09 p.m.