similarity_score: Compute the similarity from a predefined condition group

In GNET2: Constructing gene regulatory networks from expression data through functional module inference

Description

Compute the similarity between a predefined condition grouping and the sample cluster of each module, which is defined as the Adjusted Rand index between the two vectors, or the inverse of K-L divergence between the upper triangle matrix of the pairwise distance of predefined ranked condition grouping and the pairwise distance of sample cluster of each module.

Usage

similarity_score(gnet_result, group, ranked = FALSE)

Arguments

gnet_result	Returned results from gnet().
group	predefined condition grouping
ranked	the grouping information is categorical(treatment/control) or ordinal(dosage, time points)?

Value

A list of similarity scores between a predefined condition grouping and the sample cluster of each module, and the p values for the similarity scores based on permutation.

Examples

set.seed(1)
init_group_num = 8
init_method = 'kmeans'
exp_data <- matrix(rnorm(50*10),50,10)
reg_names <- paste0('TF',1:5)
rownames(exp_data) <- c(reg_names,paste0('gene',1:(nrow(exp_data)-length(reg_names))))
colnames(exp_data) <- paste0('condition_',1:ncol(exp_data))
se <- SummarizedExperiment::SummarizedExperiment(assays=list(counts=exp_data))
gnet_result <- gnet(se,reg_names,init_method,init_group_num)
s <- similarity_score(gnet_result,rep(1:5,each = 2))

GNET2 documentation built on Nov. 8, 2020, 8:03 p.m.