mergeGroups: Merge groups in a partition

In GRridge: Better prediction by use of co-data: Adaptive group-regularized ridge regression

Description

Pathway-based partition often contains a considerable number of gene sets (or groups). This function merges groups resulted by matchGeneSets. The first principal component in each group is calculated. Hierarchical clustering analysis is then performed on the first principal components from all groups. Important Note: re-grouping is only done in the non-reminder group.

Usage

mergeGroups(highdimdata, initGroups=initGroups,maxGroups=maxGroups,
                    methodDistance="manhattan", methodClust="complete")

Arguments

highdimdata	Matrix or numerical data frame. Contains the primary data of the study. Columns are samples, rows are features.
initGroups	A list of initial given groups resulted from matchGeneSets.
maxGroups	Numeric. The new desired number of groups.
methodDistance	The distance method used for clustering. See dist for futher options. Default: manhattan distance.
methodClust	The agglomeration method used for Grouping. See hclust for further options. Default: complete.

Value

A list object containing:

newGroups	A list the components of which contain the indices of the features belonging to each of the group. This object is the same as the object created by matchGeneSets and CreatePartition
newGroupMembers	A list of members in the new merged groups.

Author(s)

Putri W. Novianti

See Also

Creating partitions: CreatePartition. Creating partitions based on overlaping groups (gene sets/pathways): matchGeneSets.

Examples

# Source: http://software.broadinstitute.org/gsea/msigdb/collections.jsp
# A GMT file containing information about transcription factor targets should 
# be downloaded first from the aforementioned source.
# Section C3: motif gene sets; subsection: transcription factor targets;
# file: "c3.tft.v5.0.symbols.gmt"
# Details of the gene sets:
# Gene sets contain genes that share a transcription factor binding site 
# defined in the TRANSFAC (version 7.4, http://www.gene-regulation.com/) database.
# Each of these gene sets is annotated by a TRANSFAC record.

# Load data objects
data(dataWurdinger)

# Transform the data set to the square root scale
dataSqrtWurdinger <- sqrt(datWurdinger_BC)

#Standardize the transformed data
datStdWurdinger <- t(apply(dataSqrtWurdinger,1,function(x){(x-mean(x))/sd(x)}))

# A list of gene names in the primary RNAseq data
genesWurdinger <- as.character(annotationWurdinger$geneSymbol)

## Creating partitions based on pathways information (e.g. GSEA object)
## Some variables may belong to more than one groups (gene sets).
## The argument minlen=25 implies the minimum number of members in a gene set
## If remain=TRUE, gene sets with less than 25 members are grouped to the
## "remainder" group.
## The "TFsym" is available on https://github.com/markvdwiel/GRridgeCodata
# gseTF <- matchGeneSets(genesWurdinger,TFsym,minlen=25,remain=TRUE)

## Regrouping gene sets by hierarchical clustering analysis.
## The number of gene sets from the GSEA database is relatively too high to be used 
## in the GRridge model. Here, the initial gene sets are re-grouped into maxGroups=5, using
## information from the primary data set.
# gseTF_newGroups <- mergeGroups(highdimdata=datStdWurdinger, initGroups =gseTF, maxGroups=5);

## Extracting indices of new groups
## This following object (gseTF2) can be used further as an input
## in the "partitions" argument in the "grridge" function
# gseTF2 <- gseTF_newGroups$newGroups

## Members of the new groups
# newGroupMembers <- gseTF_newGroups$newGroupMembers

GRridge documentation built on Nov. 8, 2020, 5:47 p.m.