Brad,
Thanks for the clarification! Do you mean |30-100|/100 in your
example?
BTW, do you have a small dataset and expected output for us to test
the new feature? Thanks!
FYI, Jianhong has a lot in his plate now. So it might take some time
to get this implemented.
Best regards,
Julie
On 7/8/14 1:47 PM, "Brad Davis" <bdavis@bcgsc.ca> wrote:
Hi Julie,
The way I would calculate the percentage is for each intersection I
would divide the number of elements identified in that intersection by
the total number of elements contained in each of the included sets.
So if set 1 had 100 elements, and set 2 had 50 elements, and the
intersection has 30 elements, then the percentage of the intersection
would be 30/(100+50), while the percentage for set 1 would be
|50-100|/100 and set 2 would be |50-30|/50, that way the percentages
would represent the fraction of the elements contained in that set
that potentially could be found in that set that are found in that
set. Alternately one could calculate the percentages as the number of
elements found in that set divided by the total number of (unique)
elements found in all sets, in which case the percentage represents
the fraction of elements that are unique to that particular set. Both
would be useful, actually, but just for illustrating different things.
Brad
On 2014-07-08, 10:40 AM, "Zhu, Lihua (Julie)" <julie.zhu@umassmed.edu>
wrote:
Brad,
Thinking it over, I think it would be impossible to draw percentage
for the overlap part since it will be different for different input
peak sets unless all input peak lists have exactly the same number of
peaks or bases. Please let us know if I misunderstood you. Thanks!
Best regards,
Julie
On 7/8/14 1:33 PM, "Lihua Julie Zhu" <julie.zhu@umassmed.edu> wrote:
Brad,
Thanks for the great suggestion!
We could add two parameters: DrawPercentage (TRUE/FALSE, default
FALSE) and overalapInBaseOrPeak (peak, base), default peak. If I
understand you correctly, overlapBaseOrPeak selection only affects the
venn diagram and p-value calculation. Correct? Do you have a small
dataset and expected output for us to test the new feature? Thanks!
Best regards,
Julie
On 7/8/14 1:14 PM, "Brad Davis" <bdavis@bcgsc.ca> wrote:
Hello Jianhong and Julie,
Absolutely, thats great. I had two more suggestions for options for
makeVennDiagram that would make it even more useful to me. The first
would be to report the percentage (instead of the numbers) inside the
venn diagrams? And second, it would be useful if instead of just
reporting the total number of overlapping regions, if an option was
available to report the total amount of overlap (in terms of the
number of base pairs) instead of just the number of overlaps. That
seems like it might be particularly relevant for some of the broad
marks which can be called by some peak calling tools as a long series
of short marks as opposed to a single long contiguous region?
Thanks,
Brad
On 2014-07-08, 10:09 AM, "Zhu, Lihua (Julie)" <julie.zhu@umassmed.edu>
wrote:
Brad,
Does the following code snippet meet your needs? Thanks!
Best regards,
Julie
On 7/7/14 10:59 AM, "Umms" <jianhong.ou@umassmed.edu> wrote:
Hi Brad,
May I know what version of ChIPpeakAnno you are using? If you are
using the released version 2.12.1, you can easily change the output
filename by giving a filename for output, for example:
Library(ChIPpeakAnno)
peaks1 = RangedData(IRanges(start = c(967654, 2010897, 2496704),
end = c(967754, 2010997, 2496804), names = c("Site1", "Site2",
"Site3")),
space = c("1", "2", "3"),
strand=as.integer(1),feature=c("a","b","f"))
peaks2 = RangedData(IRanges(start = c(967659,
2010898,2496700,3075866,3123260),
end = c(967869, 2011108, 2496920, 3076166, 3123470),
names = c("t1", "t2", "t3", "t4", "t5")),
space = c("1", "2", "3", "1", "2"), strand = c(1, 1, -1,-1,1),
feature=c("a","b","c","d","a"))
pdf("path/to/AnyFileNameYouWant.pdf")
makeVennDiagram(RangedDataList(peaks1,peaks2), NameOfPeaks=c("TF1",
"TF2"),
totalTest=100,scaled=F, euler.d=F)
dev.off()
And because we already defined the the filename arguments to NULL when
calls venn.diagram, you can not define the filename by directly input
the filename to makeVennDiagram. Why we do this is because it can
output other formats of figure other than tiff. I am sorry this is not
clear in the documentation file. We will improve the documentation
later. Thank you.
Yours Sincerely,
Jianhong Ou
LRB 670A
Program in Gene Function and Expression
364 Plantation Street Worcester,
MA 01605
From: <zhu>, "Lihua (Julie)" <julie.zhu@umassmed.edu>
Date: Wednesday, July 2, 2014 8:10 PM
To: Brad Davis <bdavis@bcgsc.ca>
Cc: Jianhong Ou <jianhong.ou@umassmed.edu>,
"<bioconductor@stat.math.ethz.ch>" <bioconductor@stat.math.ethz.ch>
Subject: Re: Suggestion for a small change to ChIPpeakAnno
Brad,
Thanks for the suggestion ! We will look into it.
Best regards,
Julie
On Jul 2, 2014, at 5:55 PM, Brad Davis <bdavis@bcgsc.ca> wrote:
Suggestion for a small change to ChIPpeakAnno Hello Dr. Zhu,
I am working with your ChIPpeakAnno package for R and I find it very
convenient, but I have a small suggestion to enhance its
functionality, at least from the perspective of automating data
processing. At the moment your package creates the venn diagrams with
a defined output filename Rplots.pdf, but it would be very helpful
if the user could specify that file name in the command and overwrite
the default filename. I tried to do this by specifying filename= in
the makeVennDiagram function, but it returns an error Error in
venn.diagram(x = vennx, filename = NULL, cat.cex = 1, cat.col =
"black", :
formal argument "filename" matched by multiple actual arguments.
Thanks,
Brad
[[alternative HTML version deleted]]
