corrheatmap: Correlation heatmap for a single module

In arbet003/discoMod: Identifying differentially co-expressed modules between two groups

Description

Plot correlation heatmap of a module for both phenotype groups. ggcorrplot::ggcorrplot() is used to produce the heatmaps, which are returned in a list object in case the user wants to further customize the plots using ggplot2.

Usage

corrheatmap(
  group1_mod,
  group2_mod,
  cortype = "spearman",
  hc.order = TRUE,
  group_order = 1,
  tl.cex = 6,
  ...
)

Arguments

group1_mod	matrix of a single module in phenotype group1 (samples in rows, genes in columns)
group2_mod	matrix of the corresponding module in phenotype group1 (samples in rows, genes in columns)
cortype	"spearman" (default), "pearson", or "kendall"; type of correlation used to measure the network structure within a module.
hc.order	TRUE/FALSE, whether to use hierarchical clustering to sort genes
group_order	= 1 or 2, if hc.order=TRUE, then whether to perform the clustering in group 1 or 2. If you used find_modules() to derive the modules, then ideally, group_order should be set to the same value that you set cluster_group too in find_modules().
tl.cex	font size for variable labels (if there are many variable names, e.g. > 100, you might want to set tl.cex = 0 to just remove the variable names from the plot)
...	additional parameters passed to ggcorrplot::ggcorrplot()

Value

• both_plots : both heatmaps are combined into a single ggplot2 object using ggpubr::ggarrange()

• group1_plot : ggplot2 correlation heatmap of module in phenotype group 1

• group2_plot : ggplot2 correlation heatmap of module in phenotype group 2

Examples

######## load example dataset
library(multtest)
data(golub)
X1 = golub[,which(golub.cl==0)]
X2 = golub[,which(golub.cl==1)]
rownames(X1) = golub.gnames[,3]
rownames(X2) = golub.gnames[,3]

# use subset of 200 genes for example
set.seed(1234)
ind = sample(1:nrow(X1),200)
X1 = X1[ind,]
X2 = X2[ind,]

######## Derive modules in group 1
modules = find_modules(X1,X2,cluster_group=1)
modules$num_modules # number of modules estimated by BIC (modules with < 3 genes are excluded)
ngm = unlist(lapply(modules$group1_modules, ncol)) # number of genes per module
summary(ngm)


######## test modules for differential co-expression
testmods = test_modules(group1_modules = modules$group1_modules,group2_modules = modules$group2_modules)
# View(testmods$pvalues)
# View(testmods$qvalues)

which(testmods$pvalues$PND6 <= 0.05)
which(testmods$qvalues$PND6 <= 0.05)

# use parallel computing:
# testmods = test_modules(group1_modules = modules$group1_modules,
#                        group2_modules = modules$group2_modules,
#                        parallel=TRUE,
#                        cores=4)


######## plot module 5
heat = corrheatmap(modules$group1_modules[[5]],modules$group2_modules[[5]])
# plot(heat$both_plots)
# plot(heat$group1_plot)
# plot(heat$group2_plot)

arbet003/discoMod documentation built on Dec. 19, 2021, 4:36 a.m.