1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
|
\name{intersectDiagram}
\alias{intersectDiagram}
\title{Display set intersections}
\description{Display set intersections as rows of rectangles.}
\usage{
intersectDiagram(x,pct=FALSE,show.nulls=FALSE,xnames=NULL,namesep="+",
mar=c(0,0,3,0),main="Intersection Diagram",cex=1,col=NULL,
minspacing=NA,all.intersections=FALSE)
}
\arguments{
\item{x}{A list containing as many numeric vectors as there are sets. The
first vector contains the counts or percentages of the elements that are
only in one set, the next vector contains the counts or percentages of
elements that are in two sets and so on. A matrix of set membership
indicators or a two column matrix of object identifiers and attribute
identifiers can be passed - see Details.}
\item{pct}{Whether to display counts (FALSE) or percentages (TRUE) of the
number of entities.}
\item{show.nulls}{Whether to display the number of original objects that were
not members of any set.}
\item{xnames}{Optional user supplied names for the set categories (see
Details).}
\item{namesep}{The separator to use between category names.}
\item{mar}{The margins for the diagram. The margins that were in effect when
the function is called are restored.}
\item{main}{The title for the diagram.}
\item{col}{Colors for the sets (see Details).}
\item{cex}{Character expansion for the intersection labels.}
\item{minspacing}{The minimum spacing between the rectangles (see Details).}
\item{all.intersections}{Whether to display all intersections, even if empty
(Dangerous - see Detail).}
}
\details{
\samp{intersectDiagram} displays rows of optionally colored rectangles that
represent the intersections of set memberships (attributes) of a set of objects.
The topmost row represents the intersections of the fewest sets, and
succeeding rows represent the intersections of more sets. If there were objects
in the original data set that were not members of any set, any percentages
calculated will reflect this. By setting \samp{show.nulls} to TRUE, the counts
or percentages of such objects will be displayed below the intersections over
an empty rectangle scaled to the count or percentage.
Important - If the \samp{all.intersections} argument is TRUE, all intersections
will be displayed, whether empty or not (see the example). This is mostly for
demonstration purposes, and if the number of sets is large, is likely to produce
a very messy diagram. Similarly, sets with large numbers of
intersections that are populated will require very large displays to be readable,
even if there are small numbers in the intersections. If you would like to see
this in action, pass the data frame \samp{setdf} in the \link{categoryReshape}
example to \samp{intersectDiagram} with \samp{all.intersections} TRUE.
\samp{intersectDiagram} does not attempt to display the set intersections as
a pattern of overlapping geometric figures, but rather the relative numbers of
objects sharing each intersection. More than three intersecting sets generally
produce a complex and difficult to interpret Venn diagram, and this provides an
alternative way to display the size of intersections between larger numbers
of sets.
Each set (attribute) is assigned a color if \samp{col} is not NA. \samp{rainbow}
is called if \samp{col} is NULL, otherwise the colors passed are used. For each
intersection, the colors representing the sets intersecting are included in
the rectangle.
The strings displayed on each rectangle are taken from the argument
\samp{xnames} unless that is NULL, then the \samp{names} of the intersectList
object passed as \samp{x} or returned from the call to \samp{makeIntersectList}.
If a matrix or data frame of set membership indicators is passed as \samp{x},
it will be passed to \link{makeIntersectList} for conversion. Each column must
represent a set, and the values in the columns must be 0 or 1, or FALSE or TRUE.
Similarly, if a matrix or data frame in which the first column is object
identifiers and the second column is attributes, this will be passed to
\samp{makeIntersectList}.
The spacing between the largest rectangles is controlled by \samp{minspacing}.
\samp{minspacing} is in units of object counts and defaults to 0.1 times the
largest number of objects in an intersection. When the number of objects in
different intersections at a given level varies widely, the labels of
intersections with few objects may overlap if they are wide relative to the
rectangle representing the number of objects. This can be corrected by
passing a \samp{minspacing} argument that will increase the space between
rectangles and/or decreasing the character size of the labels. If the labels
for each set are relatively long, setting \samp{namesep="+\\n"} may help.
}
\value{
Returns the intersectionList object invisibly.
}
\keyword{misc}
\author{Jim Lemon}
\seealso{\link{makeIntersectList}, \link{getIntersectList} \link{categoryReshape}}
\examples{
# create a matrix where each row represents an element and
# a 1 (or TRUE) in each column indicates that the element is a member
# of that set.
druguse<-matrix(c(sample(c(0,1),200,TRUE),
sample(c(0,1),200,TRUE),
sample(c(0,1),200,TRUE),
sample(c(0,1),200,TRUE)),ncol=4)
colnames(druguse)<-c("Alc","Tob","THC","Amp")
druglist<-makeIntersectList(druguse)
# first display it as counts
intersectDiagram(druglist)
# then as percent with non.members, passing the initial matrix
intersectDiagram(druguse,pct=TRUE,show.nulls=TRUE)
# alter the data to have more multiple intersections
druguse[which(as.logical(druguse[,1]))[1:40],2]<-1
druguse[which(as.logical(druguse[,1]))[31:70],3]<-1
druguse[,4]<-sample(c(0,1),200,TRUE,prob=c(0.9,0.1))
intersectDiagram(druguse,col=c("gray20","gray40","gray60","gray80"),cex=0.8)
# transform the spacing - usually makes it too close, first try minspacing
intersectDiagram(druguse,col="gray",minspacing=30)
# then try cex - may need both for large differences
intersectDiagram(druguse,col="gray",cex=0.8)
# create a matrix with empty intersections
druguse<-matrix(c(sample(c(0,1),20,TRUE),
sample(c(0,1),20,TRUE),
sample(c(0,1),20,TRUE),
sample(c(0,1),20,TRUE)),ncol=4)
# show only the populated intersections
intersectDiagram(druguse)
# show all intersections
intersectDiagram(druguse,all.intersections=TRUE)
}
|
