clusterDE: Cluster Differential Expression

In farrellja/URD: URD

Description

Calculates the differentially expressed genes between each cluster and its nearest cluster (as determined by centroids in gene expression space). This can be used to fuse nearby clusters that don't have significant gene expression differences between them. This function uses the binomial differential expression test (see markersBinom.)

Usage

clusterDE(
  object,
  clustering,
  genes.dist = [e-mail],
  effect.size = log(2),
  p.thresh = 0.01,
  frac.must.express = 0.1,
  genes.de = NULL,
  verbose = T
)

Arguments

object	An URD object
clustering	(Character) Name of a clustering (i.e. a column in @group.ids)
genes.dist	(Character vector) Genes to use for calculating distance (default: variable genes, NULL is all genes)
effect.size	(Numeric) Minimum log fold-change for two genes to be considered differentially expressed
p.thresh	(Numeric) Minimum significance value for teo genes to be considered expressed
frac.must.express	(Numeric) Gene must be expressed in at least this fraction of cells in one of the two clusters to be considered.
genes.de	(Character vector) Genes to consider for differential expression (default: NULL is all genes)
verbose	(Logical) Report on progress

Value

A list with entries:

• $n.de: a data frame describing each pair of clusters tested and the number of differentially expressed genes between them

• $genes: a list of data frames for each pair of clusters with the differential gene expression test results

Examples

# Test all genes for differential expression again each cluster's nearest 
# cluster (by centroid distance in variable gene expression space)
cde <- clusterDE(object, clustering="Louvain-15", genes.dist=[e-mail], 
effect.size=log(2), p.thresh=0.01, frac.must.express=0.1, genes.de=NULL, verbose=T)

farrellja/URD documentation built on June 17, 2020, 4:48 a.m.