bnlearn - Exporting networks to DOT files

Exporting networks to DOT files

Many software packages designed to analyse and display networks can read files in the DOT format; two examples that are widely used in bioninformatics are Cytoscape (link) and Gephi (link). bnlearn can export network structures (that is, bn objects) as DOT files, making it possible to use advanced layout and display capabilities that are not available from any R package.

The relevant function for exporting is write.dot().

> library(bnlearn)
> 
> dag = random.graph(LETTERS[1:10])
> write.dot(dag, file = "dag.dot")

The DOT files contains only basic information about the network structure: its nodes and its arc set.

> cat(readLines("dag.dot"), sep = "\n")

digraph {
  "A" ;
  "B" ;
  "C" ;
  "D" ;
  "E" ;
  "F" ;
  "G" ;
  "H" ;
  "I" ;
  "J" ;
  edge [dir=forward] "A" -> "I" ;
  edge [dir=forward] "C" -> "D" ;
  edge [dir=forward] "C" -> "H" ;
  edge [dir=forward] "D" -> "E" ;
  edge [dir=forward] "E" -> "J" ;
  edge [dir=forward] "F" -> "H" ;
  edge [dir=forward] "F" -> "I" ;
  edge [dir=forward] "H" -> "I" ;
}

If all arcs in the network are undirected, the network is exported as an undirected graph (note that the keyword is graph instead of digraph).

> write.dot(skeleton(dag), file = "ug.dot")
> cat(readLines("ug.dot"), sep = "\n")

graph {
  "A" ;
  "B" ;
  "C" ;
  "D" ;
  "E" ;
  "F" ;
  "G" ;
  "H" ;
  "I" ;
  "J" ;
  edge [dir=none] "A" -- "I" ;
  edge [dir=none] "C" -- "D" ;
  edge [dir=none] "C" -- "H" ;
  edge [dir=none] "D" -- "E" ;
  edge [dir=none] "E" -- "J" ;
  edge [dir=none] "F" -- "H" ;
  edge [dir=none] "F" -- "I" ;
  edge [dir=none] "H" -- "I" ;
}

Since the DOT format has no specific keyword to denote partially directed graphs, such as those returned by cpdag() and by all constraint-based structure learning algorithms, write.dot() exports them using the digraph keyword and removing the direction from the undirected arcs.

> write.dot(set.edge(dag, from = "A", to = "G"), file = "pdag.dot")
> cat(readLines("pdag.dot"), sep = "\n")

digraph {
  "A" ;
  "B" ;
  "C" ;
  "D" ;
  "E" ;
  "F" ;
  "G" ;
  "H" ;
  "I" ;
  "J" ;
  edge [dir=none] "A" -> "G" ;
  edge [dir=forward] "A" -> "I" ;
  edge [dir=forward] "C" -> "D" ;
  edge [dir=forward] "C" -> "H" ;
  edge [dir=forward] "D" -> "E" ;
  edge [dir=forward] "E" -> "J" ;
  edge [dir=forward] "F" -> "H" ;
  edge [dir=forward] "F" -> "I" ;
  edge [dir=forward] "H" -> "I" ;
}

Last updated on Mon Aug 5 02:41:51 2024 with bnlearn 5.0 and R version 4.4.1 (2024-06-14).