findDEgenes: findDEgenes

In rsait/ORdensity: ORdensity: User-friendly R package to identify Differentially Expressed genes

Description

This function clusters the potential differentially expressed (DE) genes among them so that the real DE genes can be distinguished from the not DE genes.

Usage

findDEgenes(object, ...)

Arguments

object	An object of 'ORdensity' class
numclusters	By default NULL, it inherits from the object. Optionally, an integer number indicating number of clusters.

Details

Once the potential DE genes are identified, the real DE genes and the not real DE genes or false positives must be distinguished. Since the real DE genes must have high OR values along with low FP and dFP values, and on the contrary, the false DE genes must have low OR values but high FP and dFP values, a clustering of all the potential DE genes is carried out. The clustering is based on built-on variables OR, FP and dFP (see class ORdensity) which are scaled. The clustering algorithm is pam and by default the number of clusters in the partition is obtained by silhouette. With parameter numclusters the number of clusters in the partition can be customized.

Value

A list with k lists where k is the best number of clusters found. The clusters are ordered based on their importance according to the mean OR values of the clusters (the greater the mean OR value of the cluster the more important are the genes in the cluster). The first one is the most important, the last one the less important. Each list has elements:

• numberOfGenes: Number of genes in the cluster.

• CharacteristicsCluster: Matrix with mean values and standard deviation of variables OR, FP and dFP for each cluster.

• Genes: Identification of the genes in the cluster.

See Also

summary.ORdensity, preclusteredData, plot.ORdensity, silhouetteAnalysis, ORdensity

Examples

# Read data from 2 experimental conditions
x <- simexpr[, 3:32]
y <- simexpr[, 33:62]
EXC.1 <- as.matrix(x)
EXC.2 <- as.matrix(y)
myORdensity <- new("ORdensity", Exp_cond_1 = EXC.1, Exp_cond_2 = EXC.2)
out <- findDEgenes(myORdensity)
# For instance, characteristics of cluster1, likely composed of true DE genes 
out[[1]]

rsait/ORdensity documentation built on Nov. 5, 2019, 4:16 a.m.