lp_isomap: Link prediction with ISOMAP

In galanisl/LinkPrediction: Link prediction in complex networks

Description

Given a network of interest, it computes the likelihood score of interaction, for all disconnected node pairs, based on the embedding of the network to d-dimensional Euclidean space with ISOMAP.

Usage

lp_isomap(g, d = 2, use_weights = FALSE)

Arguments

g	igraph; The network of interest.
d	integer; The dimension of the embedding space.
use_weights	logical; Indicates whether edge weights should be used to compute shortest paths between nodes.

Value

Tibble with the following columns:

nodeA	The ID of a network node.
nodeB	The ID of a network node.
scr	The likelihood score of interaction for the node pair.

Author(s)

Gregorio Alanis-Lobato [e-mail]

References

Tenenbaum, J. B. (2000) A global geometric framework for nonlinear dimensionality reduction. Science 290:2319-2323

Kuchaiev, O. et al. (2009) Geometric de-noising of protein-protein interaction networks. PLoS Comput. Biol. 5:e1000454

Examples

# Apply the ISOMAP link predictor to the Zachary Karate Club network
iso <- lp_isomap(g = karate_club, d = 2, use_weights = FALSE)

galanisl/LinkPrediction documentation built on May 17, 2019, 12:10 p.m.