Java Web Start links
--------------------

workshop page (has all the links you will need today):

http://gaggle.systemsbiology.org/pshannon/bioc2007/


boss: http://gaggle.systemsbiology.org/2005-11/boss/boss.jnlp
name translator: http://gaggle.systemsbiology.org/nameTranslations/humanStringToGeneID.70.jnlp
DMV: http://gaggle.systemsbiology.net/2005-11/dmv.jnlp
Cytoscape 1.2, for Homo sapiens, Entrez Gene IDs:
   http://gaggle.systemsbiology.org/pshannon/cy12/blankSlate/human/cy.jnlp



R commands
----------

> library (gaggle)
> gaggleInit ()
> geese ()
> setSpecies ('Homo sapiens')

# read the expression matrix
> m = read.table ('matrix.tsv', sep='\t')

# which rows have highly variable expession?
> rows = which (apply (m, 1, function (row) IQR (row) > 5))

# what are the entrex gene ID's for these rows?
> genes = row.names (m) [rows]

# make sure that the Firegoose is registered with the boss

# set the 'target goose' -- the program which will receive the broadcast
# of these five genes.
# first: which geese are currently running and known to the boss?
> geese ()

# specify the Firefox browser with the gaggle extension, aka 'Firegoose'
> setTargetGoose ('Firegoose')


# define a convenience function to save some typing on subsequent assignment
# of the target goose

> stg = function (g) setTargetGoose (geese () [g])

# now broadcast the genes
> broadcast (genes)

# we will now make a number of manipulations in the brower, eventually resulting
# a broadcast of an expanded list of genes back to R

> from.string = getNameList ()

# take a look at the identifiers
> from.string

# in your browser, return to the workshop web page, and launch the nameTranslator
# goose.  rebroadcast identifiers from STRING, through the nameTranslator, and
# thence to R

> from.string = getNameList ()

# take a look at the identifiers.  they should now (mostly) be entrez geneID's
> from.string

# how many of are there?

> length (from.string)

# how many are in in the expression matrix?

> olength (intersect (from.string, rownames (m)))  # suggested by STRING, in our data  13

# what are the new genes, found by STRING not selected by our original stringent non-specific filter?

> new.genes = setdiff (intersect (from.string, rownames (m)), genes)

# find out the variability of these new genes

> sapply (new.genes, function (row) IQR (m [row, ]))
     3108       3112        912       7078       3109       4846        207        960        972
0.6989106  0.2881418  0.6021836  0.1142142  3.8377716  0.9327259  0.3628950  0.0000000  4.1485932 
   [1] 24


# now go back to the STRING network in your browser, and broadcast the network to R, being
# sure to broadcast through the nameTranslator

# 'network ready, node count 24, edges: 60'
> network = getNetwork () # ignore warning messages

# get a summay of the network
> network

A graphNEL graph with undirected edges
Number of Nodes = 24 
Number of Edges = 60 

# see that a url for the evidence for the associations came back with with broadcast:

> edgeData (network)[[1]][[2]]

[1] "http://string.embl.de/newstring_cgi/show_edge_data.pl?taskId=y2PwtphnXwvx&node1=322867&node2=322518"

> browseURL (edgeData (network)[[1]][[2]])

# next exercise:  launch DMV and Cytoscape, play an expression movie
# note: the geese running on your computer will probably be named slighly differently
# then those show below: version numbers ('-01', '-05') will almost certainly be
# different.  keep this in mind, and set target geese appropriately

> geese ()
[1] "Human-01"                      "DMV-01"
[3] "Network"                       "Firegoose"
[5] "STRING-GeneID, Human, v7.0     "R-05

> setTargetGoose (DMV-01)   # or stg (2)

> stg (2); showGoose ()   # bring DMV to the front
  
> broadcast (m, 'ratios')

> setTargetGoose ('Human-01') # or stg (1)

> broadcast (network)