makeDGraphFromGO: Import the GO graph as a DGraph object

In hpages/graph4: Efficient graph representation and manipulation in Bioconductor

Description

makeDGraphFromGO imports the graph of GO terms as a DGraph object.

Usage

makeDGraphFromGO(godb)

Arguments

godb	Typically the GO.db object from the GO.db package. Can be missing, in which case the GO.db package must be installed and the GO.db::GO.db object is used.

Value

A DGraph object representing the GO graph. The nodes are the GO terms and the edges are directed from child to parent.

Author(s)

Hervé Pagès

References

http://geneontology.org/

See Also

• DGraph objects.

• The GO.db package.

Examples

gograph <- makeDGraphFromGO()
gograph

nodes(gograph)  # all the GO terms

## The nodes are annotated with the name, ontology, and definition
## of each term:
mcols(nodes(gograph))

## Number of terms in each sub-ontology:
table(mcols(nodes(gograph))$ontology)

## Number of parents of each node:
out_deg <- outDegree(gograph)
table(out_deg)  # 1 node has no parents

## The GO.db package introduces an artifical term called "all"
## that is the common parent of the three sub-ontologies:
nodes(gograph)[out_deg == 0]

## To find the direct children of the "all" term, we select the edges
## that go **to** that node. This shows us the top-level term of each
## sub-ontology:
gograph[toNode(gograph) == "all"]
fromNode(gograph[toNode(gograph) == "all"])

## Note that the "all" term has no parent:
gograph[fromNode(gograph) == "all"]

## Thanks to the "all" term the graph is connected:
isConnected(gograph)

## After removing the 3 edges that go to the "all" term, the graph
## is no longer connected (it has 3 connected components):
gograph2 <- gograph[toNode(gograph) != "all"]
connComp(gograph2)

## Comparison with a graphNEL-based solution:
gognel <- as(gograph, "graphNEL")  # takes about 80 sec.
gognel

head(nodes(gognel))
head(nodes(gograph))           # nodes are annotated

object.size(gognel)            # 185.5 MB
object.size(gograph)           # 18.5 MB

deg0 <- degree(gognel)         # takes about 0.2 sec.
out_deg <- outDegree(gograph)  # takes about 0.01 sec.
in_deg <- inDegree(gograph)    # takes about 0.01 sec.
stopifnot(all.equal(deg0$outDegree, out_deg))
stopifnot(all.equal(deg0$inDegree, in_deg))


  isConnected(gognel)          # takes about 1 h
  isConnected(gograph)         # takes < 0.2 sec

hpages/graph4 documentation built on March 15, 2021, 5:17 p.m.