neat {neat}

Performs neat for lists of gene sets

Description

Compute NEAT (Signorelli et al., 2016), a test for network enrichment analysis between/from a first list of sets ('A sets') and/to a second list of sets ('B sets').

Usage

neat(alist, blist = NULL, network, nettype, nodes, 
alpha = NULL, mtc.type = 'fdr', anames = NULL, bnames = NULL)

Arguments

alist

List of A sets. Each element within the list is a vector of genes and represents a gene set

blist

List of B sets. Each element within the list is a vector of genes and represents a gene set. If nettype == "undirected", this argument is optional: if provided, every set of blist is compared with every set of alist; if NULL, the function compares sets in alist between themselves

network

One of the following objects: an adjacency matrix of class "matrix" (see 'Example 1') or a sparse adjacency matrix of class "dgCMatrix"; an igraph object (see 'Example 2'); a two-column matrix where every row represents and edge (for directed networks, parent nodes must be in the first column, and child nodes in the second)

nettype

Either 'directed' or 'undirected'

nodes

Vector containing the (ordered) names of all nodes in the network

alpha

Significance level of the test (optional). If specified, a column with the conclusion of the test is added to the output

mtc.type

Type of multiple testing correction (NB: added from package version 1.2.0). Use 'fdr' or 'BH' for the Benjamini-Hockberg method, and 'none' if no multiple testing correction is required. To know the shortcuts for other multiple testing correction methods, see p.adjust

anames

Vector of names for the elements of alist (optional: it has to be provided only if the elements of alist are not named)

bnames

Vector of names for the elements of blist (optional: it has to be provided only if the elements of blist are not named)

Value

A data frame with the following columns:

A

A set

B

B set

nab

observed number of links from A to B

expected_nab

expected number of links from A to B (in absence of enrichment)

pvalue

p-value of the test

adjusted.p

p-value adjusted to account for multiple testing

conclusion

conclusion of the test (only if alpha is specified): no enrichment, overenrichment or underenrichment

Author(s)

Mirko Signorelli

References

Signorelli, M., Vinciotti, V., Wit, E. C. (2016). NEAT: an efficient network enrichment analysis test. BMC Bioinformatics, 17:352. Url: https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-016-1203-6.

See Also

networkmatrix, plot.neat, print.neat, summary.neat

Examples

# Example 1: network given as adjacency matrix:
A = matrix(0, nrow=7, ncol=7)
A[1,c(2,3)]=1; A[2,c(5,7)]=1;A[3,c(1,4)]=1;A[4,c(2,5,7)]=1;A[6,c(2,5)]=1;A[7,4]=1
labels = letters[1:7]
set1 = c('a','e')
set2 = c('c','g')
set3 = c('d','f')
alist = list('set 1' = set1, 'set 2' = set2)
blist = list('set 3' = set3)

# test without multiple testing correction
test1 = neat(alist = alist, blist = blist, network=A, 
             nettype='directed', nodes=labels, 
             alpha=0.05, mtc.type = 'none')
print(test1)

# test with FDR multiple testing correction (default)
test1 = neat(alist = alist, blist = blist, network=A, 
             nettype='directed', nodes=labels, 
             alpha=0.05, mtc.type = 'fdr')
print(test1)

# Example 2: network given as igraph object:
library(igraph)
network = erdos.renyi.game(15, 1/3)
set1 = 1:4
set2 = c(2,5,13)
set3 = c(3,9,14)
set4 = c(8,15,20)
alist = list('set 1' = set1, 'set 2' = set2)
blist = list('set 3' = set3, 'set 4' = set4)

test2 = neat(alist, blist, network = network, 
             nettype='undirected', nodes=seq(1,15), 
             alpha=NULL)
print(test2)

# Example 3: network given as list of links:
networklist = matrix(nrow=13, ncol=2)
networklist[,1]=c('a','a','b','b','c','d','d','d','f','f','f','h','h')
networklist[,2]=c('d','e','e','g','d','b','e','g','a','b','e','c','g')

labels = letters[1:8]
set1 = c('a','b','e')
set2 = c('c','g')
set3 = c('d','f')
set4 = c('a','b','f')
alist = list('set 1' = set1, 'set 2' = set2)
blist = list('set 3' = set3, 'set4' = set4)

test3 = neat(alist, blist, network = networklist, 
             nettype = 'undirected', nodes=labels, 
             alpha=0.05, mtc.type = 'none')
print(test3)

alist = list('set 1' = set1, 'set 2' = set2, 'set 3' = set3)
test4 = neat(alist, network = networklist, 
             nettype = 'undirected', nodes=labels, 
             alpha=0.05, mtc.type = 'none')
print(test4)

# Example 4: ESR data
## Not run: 
data(yeast)
esr = list('ESR 1' = yeast$esr1, 'ESR 2' = yeast$esr2)
test = neat(alist = esr, blist = yeast$goslimproc, network = yeast$yeastnet,
            nettype = 'undirected', nodes = yeast$ynetgenes, alpha = 0.01)
# Replace with "blist = yeast$kegg" to use kegg pathways

m = dim(test)[1]
test1 = test[1:(m/2),]
table(test1$conclusion)
plot(test1)
o1=test1[test1$conclusion=='Overenrichment',]
print(o1, nrows='ALL') #display overenrichments

test2 = test[(m/2+1):m,]
table(test2$conclusion)
plot(test2)
o2=test2[test2$conclusion=='Overenrichment',]
print(o2, nrows='ALL') #display overenrichments

## End(Not run)