plotZenithResults: Heatmap of zenith results using ggplot2

In GabrielHoffman/zenith: Gene set analysis following differential expression using linear (mixed) modeling with dream

Heatmap of zenith results using ggplot2

Description

Heatmap of zenith results showing genesets that have the top and bottom t-statistics from each assay.

Usage

plotZenithResults(
  df,
  ntop = 5,
  nbottom = 5,
  label.angle = 45,
  zmax = NULL,
  transpose = FALSE,
  sortByGeneset = TRUE
)

Arguments

df	result data.frame from zenith_gsa
ntop	number of gene sets with highest t-statistic to show
nbottom	number of gene sets with lowest t-statistic to show
label.angle	angle of x-axis label
zmax	maxium of the color scales. If not specified, used range of the observed t-statistics
transpose	transpose the axes of the plot
sortByGeneset	use hierarchical clustering to sort gene sets. Default is TRUE

Value

Heatmap showing enrichment for gene sets and cell types

Examples

# Load packages
library(edgeR)
library(variancePartition)
library(tweeDEseqCountData)

# Load RNA-seq data from LCL's
data(pickrell)
geneCounts = exprs(pickrell.eset)
df_metadata = pData(pickrell.eset)

# Filter genes
# Note this is low coverage data, so just use as code example
dsgn = model.matrix(~ gender, df_metadata)
keep = filterByExpr(geneCounts, dsgn, min.count=5)

# Compute library size normalization
dge = DGEList(counts = geneCounts[keep,])
dge = calcNormFactors(dge)

# Estimate precision weights using voom
vobj = voomWithDreamWeights(dge, ~ gender, df_metadata)

# Apply dream analysis
fit = dream(vobj, ~ gender,df_metadata)
fit = eBayes(fit)

# Load Hallmark genes from MSigDB
# use gene 'SYMBOL', or 'ENSEMBL' id
# use get_GeneOntology() to load Gene Ontology
gs = get_MSigDB("H", to="ENSEMBL")
   
# Run zenith analysis
res.gsa = zenith_gsa(fit, gs, 'gendermale', progressbar=FALSE )

# Show top gene sets
head(res.gsa, 2)

# for each cell type select 3 genesets with largest t-statistic
# and 1 geneset with the lowest
# Grey boxes indicate the gene set could not be evaluted because
#    to few genes were represented
plotZenithResults(res.gsa)

GabrielHoffman/zenith documentation built on March 10, 2024, 11:43 p.m.