I'm not 100% sure that it does what you want, but when you want to do biological network visualization, the answer is often Cytoscape.

Gephi gephi.org) is a really cool and new alternative to Cytoscape - allow for dynamic manipulation of the graph, among other things. Open source, and while still alpha, very powerful. They have a demo video somewhere on the site, be sure to check it out.

They also have an API to generate graphs in multiple formats, which would allow you to directly import your node / edge data.

R / Bioconductor has 'annotation' packages that map between common identifiers and other useful entities, e.g., the KEGG.db package to manage KEGG ids and pathway names, and the org.Hs.eg.db package to manage H. sapiens annotations, including Entrez, KEGG, SYMBOL, GENENAME, etc. There is also a graph package to create and manipulate graphs, and various alternatives for visualization such as the Rgraphviz and RCytoscape packages. Here's a work flow that retrieves the MAPK and Wnt pathways, finds human Entrez gene ids associated with these pathways, translates the Entrez gene ids to gene symbols, then plots the result using Rgraphviz.

library(KEGG.db)
library(org.Hs.eg.db)
library(graph)
library(Rgraphviz) # Rgraphviz can be tricky to install, esp. on Windows

## KEGG pathways of interest
terms <- c("MAPK signaling pathway", "Wnt signaling pathway")

## map -- KEGG id links the KEGG and org.Hs.eg packages
name2id <- toTable(KEGGPATHNAME2ID[ terms ])
id2gene <- toTable(revmap(org.Hs.egPATH)[ name2id$path_id ])
id2gene$symbol <-                 # add gene SYMBOL for each Entrez id
    unlist(mget(id2gene$gene_id, org.Hs.egSYMBOL))
path2gene <- merge(name2id, id2gene, by="path_id") # 'join'

## create a graphBAM instance
df <- with(path2gene,
           data.frame(from=symbol, to=path_id, weight=1,
                      stringsAsFactors=FALSE))
gr <- graphBAM(df, edgemode="directed")

## create a (random) subgraph
set.seed(123L)
nodes <- c(sample(df$from, 25), unique(df$to))
subgr <- subGraph(nodes, gr)

## display
to <- unique(df$to)                 # 'to' nodes, ...
nodeAttrs <-                        # ...colored organge
    list(fillcolor=structure(rep("orange", length(to)), names=to))
plot(subgr,                         # display; uses Rgraphviz
    "fdp",                          # a graphviz layout
     nodeAttrs=nodeAttrs,           # node attributes
     attrs=list(                    # default attributes
       node=list(shape="ellipse",   #   nodes are ellipses...
                 width="2"),        #   ... wide enough to contain symbol ids
       edge=list(arrowsize="0.75")  #   edges are not-too-large arrows
     ))

Maybe not for the faint of heart, but hopefully showing the possibilities. There are instructions for installing R and Bioconductor packages, and many additional packages, some of which (e.g., KEGGgraph) provide a different interface; RCytoscape is interesting because the graph can be assembled programmatically, and then manipulated in Cytoscape.

You might also investigate these:

• PathCase
• Reactome
• BioCyc

I'll keep thinking.

EDIT:

• NAViGaTOR
• Panther
• CellDesigner

You could also try to download the KEGG pathways in the GPML format used by WikiPathways and PathVisio from: http://www.pathvisio.org/wiki/PathVisioDownload (look for KEGG converted pathways). And to access these in Cytoscape using the Cytoscape plugin for PathVisio (at http://www.pathvisio.org/wiki/Cytoscape_plugin ). The conversion and manual curation may have helped to get the direction in the graphs better.

Non bio specific:

arborjs Does graph visualisation in the browser with jquery/web workers.

neo4j is a graph database that can be viewed with Gephi (as with Michael's answer).

Another cool software is BioLayout Express 3d, which can make 3d representations of a network and has also been featured in a Nature Protocols cover a while ago.

For anyone stumbling on this page through google search, there is a graphite package on bioconductor which can be used to download pathways from the major databases and represent those pathways as networks.

http://www.bioconductor.org/packages/devel/bioc/html/graphite.html