Subsetting Directed igraph

Question

I'm working with a directed network in igraph. Here's some code to generate such a network:

# example graph
# install.packages(c("igraph"), dependencies = TRUE)
library(igraph)
set.seed(1)
g <- erdos.renyi.game(20, 1/20,directed=TRUE,loops=FALSE)
V(g)$name <- letters[1:20]
par(mar=rep(0,4))
plot(g)

I would like to extract subsets of this network that include an arbitrary vertex and all edges and vertices that direct to this vertex, regardless of the degree or distance of that connection.

Here's a photoshopped example of what I would like to extract, using vertex "E" in this case. I would like to extract a network that includes all vertices labeled in blue and connected edges.

Answer 1

Besides what @42-, I think distance can also be used.

> d = distances(g, to='e', mode='out')
> V(g)[which(!is.infinite(d) & d >0)]
+ 4/20 vertices, named:
[1] a n r t

Briefly, the code within the brackets returns the indices of vertices that has nonzero and finite distances from e to others.

Answer 2

It seems that the edge_connectivity funciton is a good candidate for this task. It has source and target parameters that consider the directedness of the edges:

 sapply(V(g) , function(v) if( v != 5){edge_connectivity(g, source=v, target=5)} else NA )
 a  b  c  d  e  f  g  h  i  j  k  l  m  n  o  p  q  r  s  t 
 1  0  0  0 NA  0  0  0  0  0  0  0  0  1  0  0  0  1  0  1 
 # set the retruned vector to the value named `e_con`

 #then select the vertices
 V(g)[ which(e_con >0)]
+ 4/20 vertices, named, from ba45ff0:
[1] a n r t

It's possible to recover a graph object with that list:

small_g <- delete_vertices( g, !V(g) %in% c(5, V(g)[ which(econ >0)] ))

plot( small_g)

And you asked for the edge list:

edges(small_g)
[[1]]
IGRAPH 11d63f6 DN-- 5 4 -- Erdos renyi (gnp) graph
+ attr: name (g/c), type (g/c), loops (g/l), p (g/n), name (v/c)
+ edges from 11d63f6 (vertex names):
[1] n->a t->a a->e r->e

Answer 3

You can get the neighbourhood graph of a specific node(s) using make_ego_graph. Set the order to the number of vertices (or n-1) to allow the full graph to be traversed.

make_ego_graph(g, order=length(V(g)), nodes="e", mode="in")

Answer 4

I thought all three of the answers given here were useful, but the solution from @Zhiya is the one that I ended up using (it appears to be the computationally fastest method).

That said, I did end up using elements from @42-, because my desired output was a subsetted network. Here are the changes to address this:

d = distances(g, to='e', mode='out')
vertices.to.delete <- row.names(d)[which(is.infinite(d))]
g1 <- igraph::delete.vertices(g, vertices.to.delete)