connection.matrix {QuantNorm}

Construct connection matrix for network analysis

Description

For data with known labels, this function constructs a connection matrix between unique labels, such as unique cell types. The returned matrix can be used for subject-wise network construction.

Usage

connection.matrix(mat, label, threshold = 0.15, closest = TRUE)

Arguments

mat

n*n dissimilarity (1-correlation) matrix (e.g. obtained by QuantNorm).

label

n-dimension vector for the labels of the n subjects. Replicates share the same label.

threshold

A number between 0 to 1. Two groups will be regarded as connected if average 1-correlation < threshold.

closest

True or False. Whether connect the closest group or not if the closest group cannot satisfy the threshold condition.

Value

Returns the connection matrix between unique labels.

Author(s)

Teng Fei. Email: [e-mail]

References

Fei et al (2018), Mitigating the adverse impact of batch effects in sample pattern detection, Bioinformatics, https://doi.org/10.1093/bioinformatics/bty117.

Examples

library(network); library(ggplot2); library(sna); library(GGally) #drawing network graph

data("ENCODE")

#Assigning the batches based on species
batches <- c(rep(1,13),rep(2,13))

#QuantNorm correction
corrected.distance.matrix <- QuantNorm(ENCODE,batches,method='row/column', cor_method='pearson',
                                       logdat=FALSE,standardize = TRUE, tol=1e-4)

#Constructing connection matrix

mat <- connection.matrix(mat=corrected.distance.matrix,label=colnames(corrected.distance.matrix))

#Creating network object and plot
ENCODE.net=network(mat, directed=FALSE)
ENCODE.net %v% "Species" <- c(rep('Human',13),rep('Mouse',13))
p0 <- ggnet2(ENCODE.net,label=TRUE,color = 'Species', palette = "Set2",
             size = 3, vjust = -0.6,mode = "kamadakawai",label.size = 3,
             color.legend = 'Species')+theme(legend.position = 'bottom')
plot(p0)