fetch miRNA target and map them to pathways using R/bioconductor

cited from: 
Bioconductor Digest, Vol 108, Issue 28


###################################################
here is the outline on how to do this using the miRNA predicted
conserved targets using targetscan in Human (it can also be done in Mouse).

myMir <- "hsa-mir-17"

# 1. to find targets we need to use the mature form of the mirna
# we can do this using 'mirbase.db'

require("mirbase.db") || stop("Could not load package 'mirbase.db'")

## inspect the mature forms for this particular miRNA
print(get(myMir, mirbaseMATURE))
##Accession: MIMAT0000070
##  ID: hsa-miR-17
##  Start: 14
##  End: 36
##  Evidence: experimental
##  Experiment: cloned [2,5-8], Northern [4]
##
##Accession: MIMAT0000071
##  ID: hsa-miR-17*
##  Start: 51
##  End: 72
##  Evidence: experimental
##  Experiment: cloned [1,5,7-8], Northern [1]

## select the first one
myMature <- matureName(get(myMir, mirbaseMATURE))[1]

# 2. find the targets of this mature mirna using targetscan

require("targetscan.Hs.eg.db") || stop("Could not load package",
                                       "'targetscan.Hs.eg.db'")

## get the seed-based family associated with this mature mirna
myMirFam <- get(myMature, targetscan.Hs.egMIRBASE2FAMILY)

## retrieve targets (as NCBI Gene IDs)
myMirTargets <- get(myMirFam, revmap(targetscan.Hs.egTARGETS))

##length(myMirTargets)
##[1] 1114

## 3. map target genes to KEGG pathways
require("org.Hs.eg.db") || stop("Could not load package",
                                "'org.Hs.eg.db'.")

myMirTargetsKegg <- mget(myMirTargets, org.Hs.egPATH)
##sum(!is.na(myMirTargetsKegg))
##[1] 315

## optional for convenience, add KEGG pathway names
require("KEGG.db") || stop("Could not load package 'KEGG.db'")

## using only the entries with a KEGG id
myMirTargetsKeggNames <- lapply(myMirTargetsKegg[!is.na(myMirTargetsKegg)],
                                function(i) mget(i, KEGGPATHID2NAME))


Note that since KEGG is no longer public and that the BioConductor
package will soon be considered deprecated you could use the
'reactome.db' package instead for mapping pathways. Just replace the
step 3. above by a call to:

mget(myMirTargets, reactomeEXTID2PATHID, ifnotfound=NA)

I'll leave finding the pathway names as an exercise ;-)