[BioC] Venn Diagram for 2, 3 and 4 sets (can be extended for

Matt Settles mattsettles at gmail.com
Sat Oct 6 18:00:48 CEST 2007


Deal List,

I had this problem a while back, so I had extended the limma
vennDiagram function to work with 4 (not more) using the ellipse
method some have eluded to.

An example can be found
http://bioinfo-mite.crb.wsu.edu/Rcode/pmaVenns.pdf

and the code can be found
http://bioinfo-mite.crb.wsu.edu/Rcode/Venn.R

Feel free to use and modify
Matt Settles


>Dear Steffen,
>
>In principle I'd be very happy to extend the vennDiagram() function
>in limma to more than 3 groups. So please send me your code. Your
>email seems too good to be true though, because it's geometrically
>impossible to draw Venn diagrams for 4 groups with circles. Venn
>diagrams can be done with 4 groups by going to elllipses. With more
>than 5 groups, Venn diagrams seem to me to be so complicated as to be
>perhaps no longer useful. The geometric theory of Venn diagrams has
>been worked out by the statistician AWF Edwards, see eg his book
>"Cogwheels of the Mind".

>Best wishes
>Gordon

>Date: Fri, 5 Oct 2007 10:50:23 +0200
>From: Steffen Moeller <steffen_moeller at gmx.de>
>Subject: [BioC] Venn Diagram for 2, 3 and 4 sets (can be extended for
>         more)
>To: bioconductor at stat.math.ethz.ch
>
>Dear all,
>
>I needed Venn Diagrams to accomodate 4 sets rather than the 3 that I got via
>the limma package and I could not find it elsewhere.
>
># Some test data
>A<- 1:20
>B<- 1:20
>C<- 2:20
>D<- 3:21
>
># input gathered into a list, the names
># appear in the diagram
>input<-list(a=A,b=B,c=C,d=D)
>
># calculation of values that should
># go into all the fields
>vc <- getVennCounts(input)
>
># drawing of diagram
>drawVennDiagram(vc)
>
>The diagrams look like the ones from limma, i.e., no variation of
>circle shape
>and to reflect the size of the sets. The only novelty is that is also works
>on 4 sets and it is avoiding the use of classes.
>
>What should I do with these 295 lines of code now? It does not seem
>enough for
>a package on its own. Is anybody interested? Gordon?
>
>Cheers,
>
>Steffen



More information about the Bioconductor mailing list