Entering edit mode
Loren Engrav
★
1.0k
@loren-engrav-2040
Last seen 10.3 years ago
I think my initial foray into Bioconductor and GO plotting is
finished.
Thank you all for the help. Was fun.
Code was
> rows2 <- read.delim(file="339namesColorsRowsBBtab.txt", sep="\t",
check.names = FALSE)
> colors <- as.matrix(rows2)[1,]
> nAttrs = list()
> nAttrs$fillcolor <- colors
> nAttrs$color <- colors
> bpCutLeaves <- scan(file="339namesRowBB.txt", what = "character")
Read 339 items
> bpCutLeavestree <- GOGraph(bpCutLeaves, GOBPPARENTS)
> postscript ("bpCutLeavestree100_40_4.ps", width=100, height = 40,
paper="special"); plot (bpCutLeavestree, nodeAttrs=nAttrs); dev.off()
null device
1
Then "fixed" it in Illustrator
"Final" pdf graph is up at <http: homepage.mac.com="" engrav="" menu9.html="">
and
hit File Sharing arrow. Red refers to over expression, blue to under
expression, and green when the two events collapse into one GO term.
We have 1019 differentially expressed genes, loaded them up into Bio,
postscripted out the induced BP graph, and touched it up in
Illustrator.
R and Bio are not easy. The other long thread on can documentation be
improved warrants further discussion, IMO.
The postscript that emerges is useful but not elegant. For example,
the GO
terms become the labels in "text" objects. But some are one text
object and
some are two making the file more complex than need be and tough to
manipulate. But then maybe again I do not know how to do postscript
in R.
But the large question remains, at least to me
If one has 1019 differentially expressed genes (some over expressed
and some
under) and plots the induced biological process graph with no
consideration
of enrichment, etc, just plots the induced graph...
This visualizes where in the GO tree the over and under genes are
located in
the tree. But does this "simple" plot "prove" anything?
Dunno, but it is interesting that there are three bands on the right,
red
blue red. It will be fun to see what processes are involved in these
bands.
Thank you again for the help. I will now try the same thing with MF
and
then hierarchical clustering.