diffnet: Detection of Statistically Significant Changes in Paired...

In DiffNet: Detection of Statistically Significant Changes in Complex Biological Networks

Description

Performs differential network analysis for paired biological networks to identify statistically significant changes between two graphs. Currently, the approaches available for doing this include the "closed-form", "original" (dGHD) and the "fast-approx" techniques described in the paper of Mall et al. The methods works better for large-scale complex biological networks (in pairs).

Usage

diffnet(g_A = sample_grg(6, 0.15, torus = TRUE, coords = TRUE), 
        g_B = permute(g_A, c(sample(5), 6)), p = 6, 
        threshold = 1e-50, approach = "closed-form")

Arguments

g_A	An igraph object representing graph g_A
g_B	An igraph object representing the second graph B with same number of nodes.
p	Represents the number of nodes in graph g_A which is the same as number of nodes in graph g_B
threshold	Threshold after which the "closed-form" and "fast-approx" techniques switch to use a model selection criterion similar to the "original" approach to identify statistically significant changes between two networks. By default its value is 1e-250 and a good range for this value is between 1e-50 to 1e-250.
approach	Either "closed-form"/"original"/"fast-approx". By default its "closed-form"

Value

An ordered vector representing the p-value for each node. Nodes whose p-values are less than 0.01 form the differential sub-networks in paired graphs g_A and g_B.

Author(s)

Raghvendra Mall <[e-mail]>

References

https://www.researchgate.net/publication/305355965_Detection_of_statistically_significant_network_changes_in_complex_biological_networks

Examples

library("DiffNet")
# this step is optional, it helps speed up calculations, run in parallel on 2 processors
library(doParallel)
registerDoParallel(2)
# Run the differential network analysis tehcnique on sample data
p <- diffnet()

DiffNet documentation built on May 2, 2019, 9:15 a.m.