Here's an article that just became free at PNAS
http://www.pnas.org/cgi/content/full/100/26/15428As I understand it (though I'm having trouble verifying my understanding
in detail) they looked at all the genes (what preprocessing?) for 89
fully sequenced (where's the list?) organisms (all bacteria?) and
inferred relationships among the genes based on sequence data alone.
They then projected the resulting network back into the E. coli K12
genome and used unsupervised clustering on the relationships. The
result is a graph which partitions the E. coli genome into functional
modules, which compares quite favorably to the known and annotated
metabolic modules of E. coli.
Here's a link to figure 4, which illustrates the network found
http://www.pnas.org/cgi/content/full/100/26/15428/FIG4-- rec --
Locked
