Given is a tree (igraph graph) constructed from a dataframe:
library(igraph)
library(ggraph)
#Create a symbolic edge list.
edgelist_df <- data.frame("from" = c("A", "A", "A", "B", "B", "B", "C", "D", "D", "E",
"E", "F", "G", "G", "H", "I", "I", "J", "J", "J"),
"to" = c("B", "C", "D", "E", "F", "G", "H", "I", "J", "K",
"L", "M", "N", "O", "P", "Q", "R", "S", "T", "U"))
Then create a directed graph/tree from this edgelist (edgelist_df
)
graph <- graph_from_data_frame(d = edgelist_df, directed = TRUE)
Plot the tree. Here I'm using package ggraph
with function ggraph
.
ggraph(graph = graph,
layout = 'dendrogram',
circular = FALSE)
geom_edge_diagonal()
geom_node_point()
geom_node_text(aes(label = name),
angle = 0,
hjust = 1.5,
nudge_y = 0,
size = 5)
theme_void()
The question is how to return a character vector containing the names of all leaf nodes from a subtree that is specified by one node, representing the root node of that subtree. For example:
- If node = "B", then all leaf nodes that are part of the subtree with root "B" are: "K", "L", "M", "N" and "O".
- If node = "H", then all leaf nodes that are part of the subtree with root "H" are: "P".
- If node = "A", then all leaf nodes that are part of the subtree with root "A" (which is the original tree) are: "K", "L", "M", "N", "O", "P", "Q", "R", "S", "T" and "U".
CodePudding user response:
Assuming that object graph
is indeed a tree, then the function below gives the desired result.
determine_leaf_nodes_subtree <- function(graph, vertex){
#Determine the name of the root vertex.
root <- V(graph)$name[degree(graph = graph, v = V(graph), mode = "in") == 0]
#Determine the name(s) of the leaf vertex/vertices.
leafs <- V(graph)$name[degree(graph = graph, v = V(graph), mode = "out") == 0]
#Calculate the tree depth. That is, the smallest path length of all shortest paths from root
#to leaf vertices.
sh_paths <- shortest_paths(graph = graph, from = root, to = leafs)$vpath
max_sh_path_length <- max(sapply(X = sh_paths, FUN = length))
#If 'vertex' is a leaf node itself, then return the name of that vertex.
if(vertex %in% leafs){
return(vertex)
} else {
#If 'vertex' is not a leaf node, then determine the subset of all vertices that are in the
#neighborhood of 'vertex', excluding 'vertex' itself ('mindist' = 1). The maximum order or
#'depth' is specified by ('max_sh_path_length')
vertices_subset <- neighborhood(graph = graph,
order = max_sh_path_length - 1,
nodes = V(graph)[V(graph)$name == vertex],
mode = "out",
mindist = 1)
#Extract the names of the vertices in 'vertices_subset'
vertices_subset_names <- names(unlist(vertices_subset))
#The overlap/intersection of vertex names between 'vertices_subset_names' and 'leafs' gives
#all leaf nodes that are part of the subtree with root vertex 'vertex'.
result <- intersect(x = vertices_subset_names, leafs)
return(result)
}
}
With respect to the examples stated in the question, this function gives the following output:
determine_leaf_nodes_subtree(graph = graph, vertex = "B")
[1] "K" "L" "M" "N" "O"
determine_leaf_nodes_subtree(graph = graph, vertex = "H")
[1] "P"
determine_leaf_nodes_subtree(graph = graph, vertex = "A")
[1] "K" "L" "M" "N" "O" "P" "Q" "R" "S" "T" "U"
CodePudding user response:
You can make two calls to ego
: the first to grab all the nodes that can be reached from parent to child (mode="out"
), and the second to see if these selected nodes have any children (if not then they are leaf nodes.
fun <- function(graph, node="B"){
path <- ego(graph, order=length(V(graph)), nodes=node, mode="out")
nms <- names(path[[1]])
nms[ego_size(graph, order=1, nodes=nms, mode="out", mindist=1) == 0]
}
This produces
fun(graph, "B")
# [1] "K" "L" "M" "N" "O"
fun(graph, "H")
# [1] "P"
fun(graph, "A")
# [1] "K" "L" "M" "N" "O" "P" "Q" "R" "S" "T" "U"
In fact, taking a lead from Peters answer the call to ego_size
could be replaced with
nms[degree(graph, v=nms, mode="out") == 0]
CodePudding user response:
You can define a function f
with distances()
and degree
like below
f <- function(g, r) {
names(V(g))[is.finite(distances(g, r, mode = "out")) & degree(g) == 1]
}
which gives
> f(g, "B")
[1] "K" "L" "M" "N" "O"