Version: 1.63-1
Date: 2016-03-10
Title: Methods for Detection of Clusters in Hierarchical Clustering Dendrograms
Author: Peter Langfelder <Peter.Langfelder@gmail.com> and Bin Zhang <binzhang.ucla@gmail.com>, with contributions from Steve Horvath <SHorvath@mednet.ucla.edu>
Maintainer: Peter Langfelder <Peter.Langfelder@gmail.com>
Depends: R (≥ 2.3.0), stats
ZipData: no
License: GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
Description: Contains methods for detection of clusters in hierarchical clustering dendrograms.
URL: http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/BranchCutting/
NeedsCompilation: no
Packaged: 2016-03-10 18:25:27 UTC; plangfelder
Repository: CRAN
Date/Publication: 2016-03-11 00:39:02

Methods for Detection of Clusters in Hierarchical Clustering Dendrograms

Description

Contains methods for detection of clusters in hierarchical clustering dendrograms.

Details

Package: dynamicTreeCut
Version: 1.63-1
Date: 2016-03-10
Depends: R, stats
ZipData: no
License: GPL version 2 or newer
URL: http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/BranchCutting/

Index: 

cutreeDynamic           Adaptive branch pruning of hierarchical
                        clustering dendrograms.
cutreeDynamicTree       Dynamic dendrogram pruning based on dendrogram
                        only
cutreeHybrid            Hybrid adaptive tree cut for hierarchical
                        clustering dendrograms.
indentSpaces            Spaces for indented output.
merge2Clusters          Merge two clusters
printFlush              Print arguments and flush the console.
treecut-package         Methods for detection of clusters in
                        hierarchical clustering dendrograms.

Author(s)

Peter Langfelder <Peter.Langfelder@gmail.com> and Bin Zhang <binzhang.ucla@gmail.com>, with contributions from Steve Horvath <SHorvath@mednet.ucla.edu>

Maintainer: Peter Langfelder <Peter.Langfelder@gmail.com>

Adaptive Branch Pruning of Hierarchical Clustering Dendrograms

Description

This wrapper provides a common access point for two methods of adaptive branch pruning of hierarchical clustering dendrograms.

Usage

cutreeDynamic(
      dendro, cutHeight = NULL, minClusterSize = 20,

      # Basic tree cut options
      method = "hybrid", distM = NULL,
      deepSplit = (ifelse(method=="hybrid", 1, FALSE)),

      # Advanced options
      maxCoreScatter = NULL, minGap = NULL,
      maxAbsCoreScatter = NULL, minAbsGap = NULL,

      minSplitHeight = NULL, minAbsSplitHeight = NULL,

      # External (user-supplied) measure of branch split
      externalBranchSplitFnc = NULL, minExternalSplit = NULL,
      externalSplitOptions = list(),
      externalSplitFncNeedsDistance = NULL,
      assumeSimpleExternalSpecification = TRUE,

      # PAM stage options
      pamStage = TRUE, pamRespectsDendro = TRUE,
      useMedoids = FALSE, maxDistToLabel = NULL,
      maxPamDist = cutHeight,
      respectSmallClusters = TRUE,

      # Various options
      verbose = 2, indent = 0)

Arguments

dendro

A hierarchical clustering dendorgram such as one returned by hclust.

cutHeight

Maximum joining heights that will be considered. For method=="tree" it defaults to 0.99. For method=="hybrid" it defaults to 99% of the range between the 5th percentile and the maximum of the joining heights on the dendrogram.

minClusterSize

Minimum cluster size.

method

Chooses the method to use. Recognized values are "hybrid" and "tree".

distM

Only used for method "hybrid". The distance matrix used as input to hclust. If not given and method == "hybrid", the function will issue a warning and default to method = "tree".

deepSplit

For method "hybrid", can be either logical or integer in the range 0 to 4. For method "tree", must be logical. In both cases, provides a rough control over sensitivity to cluster splitting. The higher the value (or if TRUE), the more and smaller clusters will be produced. For the "hybrid" method, a finer control can be achieved via maxCoreScatter and minGap below.

maxCoreScatter

Only used for method "hybrid". Maximum scatter of the core for a branch to be a cluster, given as the fraction of cutHeight relative to the 5th percentile of joining heights. See Details.

minGap

Only used for method "hybrid". Minimum cluster gap given as the fraction of the difference between cutHeight and the 5th percentile of joining heights.

maxAbsCoreScatter

Only used for method "hybrid". Maximum scatter of the core for a branch to be a cluster given as absolute heights. If given, overrides maxCoreScatter.

minAbsGap

Only used for method "hybrid". Minimum cluster gap given as absolute height difference. If given, overrides minGap.

minSplitHeight

Minimum split height given as the fraction of the difference between cutHeight and the 5th percentile of joining heights. Branches merging below this height will automatically be merged. Defaults to zero but is used only if minAbsSplitHeight below is NULL.

minAbsSplitHeight

Minimum split height given as an absolute height. Branches merging below this height will automatically be merged. If not given (default), will be determined from minSplitHeight above.

externalBranchSplitFnc

Optional function to evaluate split (dissimilarity) between two branches. Either a single function or a list in which each component is a function (see assumeSimpleExternalSpecification below for how to specify a single function). Each function can be specified by name (a character string) or the actual function object. Each given function must take arguments branch1 and branch2 that specify the indices of objects in the two branches whose dissimilarity is to be evaluated, and possibly other arguments. It must return a number that quantifies the dissimilarity of the two branches. The returned value will be compared to minExternalSplit (see below). This argument is only used for method "hybrid".

minExternalSplit

Thresholds to decide whether two branches should be merged. It should be a numeric vector of the same length as the number of functions in externalBranchSplitFnc above. Only used for method "hybrid".

externalSplitOptions

Further arguments to function externalBranchSplitFnc. If only one external function is specified in externalBranchSplitFnc above, externalSplitOptions can be a named list of arguments or a list with one component that is in turn the named list of further arguments for externalBranchSplitFnc[[1]]. The argument assumeSimpleExternalSpecification controls which of the two possibilities should be assumed. If multiple functions are specified in externalBranchSplitFnc, externalSplitOptions must be a list in which each component is a named list giving the further arguments for the corresponding function in externalBranchSplitFnc. Only used for method "hybrid".

externalSplitFncNeedsDistance

Optional specification of whether the external branch split functions need the distance matrix as one of their arguments. Either NULL or a logical vector with one element per branch split function that specifies whether the corresponding branch split function expects the distance matrix as one of its arguments. The default NULL is interpreted as a vector of TRUE. When dealing with a large number of objects, setting this argument to FALSE whenever possible can prevent unnecessary memory utilization.

assumeSimpleExternalSpecification

Logical: when minExternalSplit above is a scalar (has length 1), should the function assume a simple specification of externalBranchSplitFnc and externalSplitOptions? If TRUE, externalBranchSplitFnc is taken as the function specification and externalSplitOptions the named list of options. This is suitable for simple direct calls of this function. If FALSE, externalBranchSplitFnc is assumed to be a list with a single component which specifies the function, and externalSplitOptions is a list with one component that is in turn the named list of further arguments for externalBranchSplitFnc[[1]].

pamStage

Only used for method "hybrid". If TRUE, the second (PAM-like) stage will be performed.

pamRespectsDendro

Logical, only used for method "hybrid". If TRUE, the PAM stage will respect the dendrogram in the sense that objects and small clusters will only be assigned to clusters that belong to the same branch that the objects or small clusters being assigned belong to.

useMedoids

Only used for method "hybrid" and only if labelUnlabeled==TRUE. If TRUE, the second stage will be use object to medoid distance; if FALSE, it will use average object to cluster distance. The default (FALSE) is recommended.

maxDistToLabel

Deprecated, use maxPamDist instead. Only used for method "hybrid" and only if labelUnlabeled==TRUE. Maximum object distance to closest cluster that will result in the object assigned to that cluster.

maxPamDist

Only used for method "hybrid" and only if labelUnlabeled==TRUE. Maximum object distance to closest cluster that will result in the object assigned to that cluster. Defaults to cutHeight.

respectSmallClusters

Only used for method "hybrid" and only if labelUnlabeled==TRUE. If TRUE, branches that failed to be clusters in stage 1 only because of insufficient size will be assigned together in stage 2. If FALSE, all objects will be assigned individually.

verbose

Controls the verbosity of the output. 0 will make the function completely quiet, values up to 4 gradually increase verbosity.

indent

Controls indentation of printed messages (see verbose above). Each unit adds two spaces before printed messages; useful when several functions' output is to be nested.

Details

This is a wrapper for two related but different methods for cluster detection in hierarchical clustering dendrograms.

In order to make the shape parameters maxCoreScatter and minGap more universal, their values are interpreted relative to cutHeight and the 5th percetile of the merging heights (we arbitrarily chose the 5th percetile rather than the minimum for reasons of stability). Thus, the absolute maximum allowable core scatter is calculated as maxCoreScatter * (cutHeight - refHeight) + refHeight and the absolute minimum allowable gap as minGap * (cutHeight - refHeight), where refHeight is the 5th percentile of the merging heights.

Value

A vector of numerical labels giving assignment of objects to modules. Unassigned objects are labeled 0, the largest module has label 1, next largest 2 etc.

Author(s)

Peter Langfelder, Peter.Langfelder@gmail.com

References

Langfelder P, Zhang B, Horvath S, 2007. http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/BranchCutting

See Also

hclust, cutreeHybrid, cutreeDynamicTree.

Dynamic Dendrogram Pruning Based on Dendrogram Only

Description

Detect clusters in a hierarchical dendrogram using a variable cut height approach. Uses only the information in the dendrogram itself is used (which may give incorrect assignment for outlying objects).

Usage

cutreeDynamicTree(dendro, maxTreeHeight = 1, deepSplit = TRUE, minModuleSize = 50)

Arguments

dendro

Hierarchical clustering dendrogram such produced by hclust.

maxTreeHeight

Maximum joining height of objects to be considered part of clusters.

deepSplit

If TRUE, method will favor sensitivity and produce more smaller clusters. When FALSE, there will be fewer bigger clusters.

minModuleSize

Minimum module size. Branches containing fewer than minModuleSize objects will be left unlabeled.

Details

A variable height branch pruning technique for dendrograms produced by hierarchical clustering. Initially, branches are cut off at the height maxTreeHeight; the resulting clusters are then examined for substructure and if subclusters are detected, they are assigned separate labels. Subclusters are detected by structure and are required to have a minimum of minModuleSize objects on them to be assigned a separate label. A rough degree of control over what it means to be a subcluster is implemented by the parameter deepSplit.

Value

A vector of numerical labels giving assignment of objects to modules. Unassigned objects are labeled 0, the largest module has label 1, next largest 2 etc.

Author(s)

Bin Zhang, binzhang.ucla@gmail.com, with contributions by Peter Langfelder, Peter.Langfelder@gmail.com.

References

http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/BranchCutting

See Also

hclust, cutreeHybrid

Hybrid Adaptive Tree Cut for Hierarchical Clustering Dendrograms

Description

Detect clusters in a dendorgram produced by the function hclust.

Usage

cutreeHybrid(
      # Input data: basic tree cutiing
      dendro, distM,

      # Branch cut criteria and options
      cutHeight = NULL, minClusterSize = 20, deepSplit = 1,

      # Advanced options
      maxCoreScatter = NULL, minGap = NULL,
      maxAbsCoreScatter = NULL, minAbsGap = NULL,

      minSplitHeight = NULL, minAbsSplitHeight = NULL,

      # External (user-supplied) measure of branch split
      externalBranchSplitFnc = NULL, minExternalSplit = NULL,
      externalSplitOptions = list(),
      externalSplitFncNeedsDistance = NULL,
      assumeSimpleExternalSpecification = TRUE,


      # PAM stage options
      pamStage = TRUE, pamRespectsDendro = TRUE,
      useMedoids = FALSE,
      maxPamDist = cutHeight,
      respectSmallClusters = TRUE,

      # Various options
      verbose = 2, indent = 0)

Arguments

dendro

a hierarchical clustering dendorgram such as one returned by hclust.

distM

Distance matrix that was used as input to hclust.

cutHeight

Maximum joining heights that will be considered. It defaults to 99 of the range between the 5th percentile and the maximum of the joining heights on the dendrogram.

minClusterSize

Minimum cluster size.

deepSplit

Either logical or integer in the range 0 to 4. Provides a rough control over sensitivity to cluster splitting. The higher the value, the more and smaller clusters will be produced. A finer control can be achieved via maxBranchCor, minBranchSplit, maxCoreScatter and minGap below.

maxCoreScatter

Maximum scatter of the core for a branch to be a cluster, given as the fraction of cutHeight relative to the 5th percentile of joining heights. See Details.

minGap

Minimum cluster gap given as the fraction of the difference between cutHeight and the 5th percentile of joining heights.

maxAbsCoreScatter

Maximum scatter of the core for a branch to be a cluster given as absolute heights. If given, overrides maxCoreScatter.

minAbsGap

Minimum cluster gap given as absolute height difference. If given, overrides minGap.

minSplitHeight

Minimum split height given as the fraction of the difference between cutHeight and the 5th percentile of joining heights. Branches merging below this height will automatically be merged. Defaults to zero but is used only if minAbsSplitHeight below is NULL.

minAbsSplitHeight

Minimum split height given as an absolute height. Branches merging below this height will automatically be merged. If not given (default), will be determined from minSplitHeight above.

externalBranchSplitFnc

Optional function to evaluate split (dissimilarity) between two branches. Either a single function or a list in which each component is a function (see assumeSimpleExternalSpecification below for how to specify a single function). Each function can be specified by name (a character string) or the actual function object. Each given function must take arguments branch1 and branch2 that specify the indices of objects in the two branches whose dissimilarity is to be evaluated, and possibly other arguments. It must return a number that quantifies the dissimilarity of the two branches. The returned value will be compared to minExternalSplit (see below). This argument is only used for method "hybrid".

minExternalSplit

Thresholds to decide whether two branches should be merged. It should be a numeric vector of the same length as the number of functions in externalBranchSplitFnc above. Only used for method "hybrid".

externalSplitOptions

Further arguments to function externalBranchSplitFnc. If only one external function is specified in externalBranchSplitFnc above, externalSplitOptions can be a named list of arguments or a list with one component that is in turn the named list of further arguments for externalBranchSplitFnc[[1]]. The argument assumeSimpleExternalSpecification controls which of the two possibilities should be assumed. If multiple functions are specified in externalBranchSplitFnc, externalSplitOptions must be a list in which each component is a named list giving the further arguments for the corresponding function in externalBranchSplitFnc. Only used for method "hybrid".

externalSplitFncNeedsDistance

Optional specification of whether the external branch split functions need the distance matrix as one of their arguments. Either NULL or a logical vector with one element per branch split function that specifies whether the corresponding branch split function expects the distance matrix as one of its arguments. The default NULL is interpreted as a vector of TRUE. When dealing with a large number of objects, setting this argument to FALSE whenever possible can prevent unnecessary memory utilization.

assumeSimpleExternalSpecification

Logical: when minExternalSplit above is a scalar (has length 1), should the function assume a simple specification of externalBranchSplitFnc and externalSplitOptions? If TRUE, externalBranchSplitFnc is taken as the function specification and externalSplitOptions the named list of options. This is suitable for simple direct calls of this function. If FALSE, externalBranchSplitFnc is assumed to be a list with a single component which specifies the function, and externalSplitOptions is a list with one component that is in turn the named list of further arguments for externalBranchSplitFnc[[1]].

pamStage

Logical, only used for method "hybrid". If TRUE, the second (PAM-like) stage will be performed.

pamRespectsDendro

Logical, only used for method "hybrid". If TRUE, the PAM stage will respect the dendrogram in the sense an object can be PAM-assigned only to clusters that lie below it on the branch that the object is merged into. See cutreeDynamic for more details.

useMedoids

if TRUE, the second stage will be use object to medoid distance; if FALSE, it will use average object to cluster distance. The default (FALSE) is recommended.

maxPamDist

Maximum object distance to closest cluster that will result in the object assigned to that cluster. Defaults to cutHeight.

respectSmallClusters

If TRUE, branches that failed to be clusters in stage 1 only because of insufficient size will be assigned together in stage 2. If FALSE, all objects will be assigned individually.

verbose

Controls the verbosity of the output. 0 will make the function completely quiet, values up to 4 gradually increase verbosity.

indent

Controls indentation of printed messages (see verbose above). Each unit adds two spaces before printed messages; useful when several functions' output is to be nested.

Details

The function detects clusters in a hierarchical dendrogram based on the shape of branches on the dendrogram. For details on the method, see http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/BranchCutting.

In order to make the shape parameters maxCoreScatter and minGap more universal, their values are interpreted relative to cutHeight and the 5th percetile of the merging heights (we arbitrarily chose the 5th percetile rather than the minimum for reasons of stability). Thus, the absolute maximum allowable core scatter is calculated as maxCoreScatter * (cutHeight - refHeight) + refHeight and the absolute minimum allowable gap as minGap * (cutHeight - refHeight), where refHeight is the 5th percentile of the merging heights.

Value

A list containg the following elements:

labels

Numerical labels of clusters, with 0 meaning unassigned, label 1 labeling the largest cluster etc.

cores

Numerical labels indicating cores of found clusters.

smallLabels

Numerical labels for branches that failed to be recognized clusters only because of insufficient number of objects.

mergeDiagnostics

A data.frame with one row per merge in the input dendrogram. The columns give the values of the various merging criteria used by the algorithm. Missing data indicate that at least one of the "branches" merged was actually a singleton (single node) and hence the branch merging was automatic.

mergeCriteria

Values of the merging thresholds. Either a copy of the corresponding input thresholds or values determined by deepSplit.

branches

A list detailing the deteced branch structure.

Author(s)

Peter Langfelder, Peter.Langfelder@gmail.com

References

Langfelder P, Zhang B, Horvath S, 2007. http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork/BranchCutting

See Also

hclust, as.dist

Spaces for Indented Output

Description

Returns a character string containing two times indent spaces.

Usage

indentSpaces(indent = 0)

Arguments

indent

Desired level of indentation. The number of returned spaces will be twice this argument.

Value

A character string containing spaces, of length twice indent.

Author(s)

Peter Langfelder, Peter.Langfelder@gmail.com

Examples

spaces = indentSpaces(0);
print(paste(spaces, "This output is not indented..."));
spaces = indentSpaces(1);
print(paste(spaces, "...while this one is."))

Merge Two Clusters

Description

Merge 2 clusters into 1.

Usage

merge2Clusters(labels, mainClusterLabel, minorClusterLabel)

Arguments

labels

a vector or factor giving the cluster labels

mainClusterLabel

label of the first merged cluster. The merged cluster will have this label.

minorClusterLabel

label of the second merged cluster.

Value

A vector or factor of the merged labels.

Author(s)

Bin Zhang and Peter Langfelder

Examples


options(stringsAsFactors = FALSE);

# Works with character labels:
labels = c(rep("grey", 5), rep("blue", 2), rep("red", 3))
merge2Clusters(labels, "blue", "red")

# Works with factor labels:
labelsF = factor(labels)
merge2Clusters(labelsF, "blue", "red")

# Works also with numeric labels:

labelsN = as.numeric(factor(labels))
labelsN
merge2Clusters(labelsF, 1, 3)

Print Arguments and Flush the Console

Description

Passes all its arguments unchaged to the standard print function; after the execution of print it flushes the console, if possible.

Usage

printFlush(...)

Arguments

...

Arguments to be passed to the standard print function.

Details

Passes all its arguments unchaged to the standard print function; after the execution of print it flushes the console, if possible.

Value

Returns the value of the print function.

Author(s)

Peter Langfelder, Peter.Langfelder@gmail.com

See Also

print