This function produces a pairwise LD plot.

LDheatmap() is used to produce a graphical display, as a heat map, of pairwise linkage disequilibrium (LD) measurements for SNPs. The heat map is a false color image in the upper-left diagonal of a square plot. Optionally, a line parallel to the diagonal of the image indicating the physical or genetic map positions of the SNPs may be added, along with text reporting the total length of the genomic region considered.

LDheatmap(gdat, genetic.distances=NULL, distances="physical",
LDmeasure="r", title="Pairwise LD", add.map=TRUE, add.key=TRUE,
geneMapLocation=0.15, geneMapLabelX=NULL, geneMapLabelY=NULL,
SNP.name=NULL, color=NULL, newpage=TRUE,
name="ldheatmap", vp.name=NULL, pop=FALSE, flip=NULL, text=FALSE)

SNP data: a data frame of genotype objects, a SnpMatrix object, a square matrix of pairwise linkage disequilibrium measurements or an object of class "LDheatmap" (the returned object of this function).


A numeric vector of map locations of the SNPs, in the same order as SNPs listed in gdat, in terms of genetic or physical distances. Physical distances should be in bases, genetic distances should be in centiMorgans (cM). When gdat is not an object of class LDheatmap, the default is a vector that represents equi-spaced markers, 1kb (1000 bases) apart. When gdat is an object of class LDheatmap, the genetic.distances argument is taken to be the genetic.distances list item of gdat.


A character string to specify whether the provided map locations are in physical or genetic distances. If distances="physical" (default), the text describing the total length of the region will be “Physical Length:XXkb” where XX is the length of the region in kilobases. If distances="genetic", the text will be “Genetic Map Length:YYcM” where YY is the length of the region in centiMorgans. If gdat is an object of class LDheatmap, distances is taken from gdat.


A character string specifying the measure of LD - either allelic correlation \(r^2\) or Lewontin's |D\('\)|; default = "r" for \(r^2\); type "D'" for |D\('\)|. This argument is ignored when the user has already supplied calculated LD measurements through gdat (i.e., when gdat is a matrix of pairwise LD measurements or an object of class "LDheatmap").


A character string for the main title of the plot. Default is “Pairwise LD”.


If TRUE (default) a diagonal line indicating the physical or genetic map positions of the SNPs will be added to the plot, along with text indicating the total length of the genetic region.


If TRUE (default) the color legend is drawn.


A numeric value specifying the position of the line parallel to the diagonal of the matrix; the larger the value, the farther it lies from the matrix diagonal. Ignored when add.map=FALSE.


A numeric value specifying the x-coordinate of the text indicating the total length of the genomic region being considered. Ignored when add.map=FALSE.


A numeric value specifying the y-coordinate of the text indicating the total length of the genomic region being considered. Ignored when add.map=FALSE.


A vector of character string(s) of SNP name(s) to be labelled. Should match the names of SNPs in the provided object gdat, otherwise nothing is done.


A range of colors to be used for drawing the heat map. Default is grey.colors(20).


If TRUE (default), the heat map will be drawn on a new page.


A character string specifying the name of the LDheatmap graphical object (grob) to be produced.


A character string specifying the name of the viewport where the heat map is going to be drawn.


If TRUE, the viewport where the heat map is drawn is popped (i.e. removed) from the viewport tree after drawing. Default=FALSE.


If TRUE, the LDheatmap plot is flipped below a horizontal line, in the style of Haploview. Default is FALSE.


If TRUE, the LD measurements are printed on each cell.


The input object gdat can be a data frame of genotype objects (a data structure from the genetics package), a SnpMatrix object (a data structure from the snpStats package), or any square matrix with values between 0 and 1 inclusive. LD computation is much faster for SnpMatrix objects than for genotype objects. In the case of a matrix of LD values between 0 and 1, the values above the diagonal will be plotted. In the display of LD, SNPs appear in the order supplied by the user as the horizontal and vertical coordinates are increased and one moves along the off-diagonal line, from the bottom-left to the top-right corner. To achieve this, the conventions of the image() function have been adopted, in which horizontal coordinates correspond to the rows of the matrix and vertical coordinates correspond to columns, and vertical coordinates are indexed in increasing order from bottom to top. For the argument color, an appropriate color palette for quantitative data is recommended, as outlined in the help page of the brewer.pal() function of the RColorBrewer package. See the package vignette LDheatmap for more examples and details of the implementation. Examples of adding ``tracks'' of genomic annotation above a flipped heatmap are in the package vignette addTracks.


An object of class "LDheatmap" which contains the following components:


The matrix of pairwise LD measurements plotted in the heat map.


A grid graphical object (grob) representing the produced heat map.


The viewport in which the heat map is drawn. See viewport.


The vector of the supplied physical or genetic map locations, or the vector of equispaced marker distances when no distance vector is supplied.


A character string specifying whether the provided map distances are physical or genetic.


The range of colors used for drawing the heat map.

The grob LDheatmapGrob has three grobs as its children (components). They are listed below along with their own children and respectively represent the color image with main title, genetic map and color key of the heat map: "heatMap" - "heatmap", "title"; "geneMap" - "diagonal", "segments", "title", "symbols", "SNPnames"; and "Key" - "colorKey", "title", "labels", "ticks", "box".


The produced heat map can be modified in two ways. First, it is possible to edit interactively the grob components of the heat map, by using the function grid.edit; the function will not work if there is no open graphical device showing the heat map. Alternatively, the user can use the function editGrob and work with the grob LDheatmapGrob returned by LDheatmap. See Examples for usage. LDheatmap() uses Grid, which does not respond to par() settings. Hence modifying par() settings of mfrow and mfcol will not work with LDheatmap(). The Examples section shows how to display multiple heat maps on one plot without the use of par().


Shin J-H, Blay S, McNeney B and Graham J (2006). LDheatmap: An R Function for Graphical Display of Pairwise Linkage Disequilibria Between Single Nucleotide Polymorphisms. Journal of Statistical Software, 16 Code Snippet 3

See Also

LD, genotype, Grid, LDheatmap.highlight, LDheatmap.marks

  • LDheatmap
# Pass LDheatmap a SnpMatrix object
#make an example matrix of genotypes, coded as 0, 1 2 copies of an index allele
#Load the package's data set
#Creates a data frame "CEUSNP" of genotype data and a vector "CEUDist"
#of physical locations of the SNPs
# Produce a heat map in a grey color scheme
MyHeatmap <- LDheatmap(CEUSNP, genetic.distances = CEUDist,
                      color = grey.colors(20))
# Same heatmap, flipped below a horizontal gene map -- for examples of
# adding genomic annotation tracks to a flipped heatmap see
# vignette("addTracks")
# flippedHeatmap<-LDheatmap(MyHeatmap,flip=TRUE)
# Prompt the user before starting a new page of graphics output
# and save the original prompt settings in old.prompt.
old.prompt <- devAskNewPage(ask = TRUE)
# Highlight a certain LD block of interest:
LDheatmap.highlight(MyHeatmap, i = 3, j = 8, col = "black", 
fill = "grey",flipOutline=FALSE, crissCross=FALSE)
# Plot a symbol in the center of the pixel which represents LD between
# the fourth and seventh SNPs:
LDheatmap.marks(MyHeatmap,  4,  7,  gp=grid::gpar(cex=2),  pch = "*")
#### Use an RGB pallete for the color scheme ####
rgb.palette <- colorRampPalette(rev(c("blue", "orange", "red")), space = "rgb")
LDheatmap(MyHeatmap, color=rgb.palette(18))
#### Modify the plot by using 'grid.edit' function ####
#Draw a heat map where the SNPs "rs2283092" and "rs6979287" are labelled.
LDheatmap(MyHeatmap, SNP.name = c("rs2283092", "rs6979287"))
#Find the names of the top-level graphical objects (grobs) on the current display
#[1] "ldheatmap"
# Find the names of the component grobs of "ldheatmap"
#[1] "heatMap" "geneMap" "Key"
#Find the names of the component grobs of heatMap
#[1] "heatmap" "title"
#Find the names of the component grobs of geneMap
#[1] "diagonal" "segments" "title"    "symbols"  "SNPnames"
#Find the names of the component grobs of Key
#[1] "colorKey" "title"    "labels"   "ticks"    "box"
#Change the plotting symbols that identify SNPs rs2283092 and rs6979287
#on the plot to bullets
grid.edit("symbols", pch = 20, gp = gpar(cex = 1))
#Change the color of the main title
grid.edit(gPath("ldheatmap", "heatMap", "title"), gp = gpar(col = "red"))
#Change size of SNP labels
grid.edit(gPath("ldheatmap", "geneMap","SNPnames"), gp = gpar(cex=1.5))
#Add a grid of white lines to the plot to separate pairwise LD measures
grid.edit(gPath("ldheatmap", "heatMap", "heatmap"), gp = gpar(col = "white",
                                                             lwd = 2))
#### Modify a heat map using 'editGrob' function ####
MyHeatmap <- LDheatmap(MyHeatmap, color = grey.colors(20))
new.grob <- editGrob(MyHeatmap$LDheatmapGrob, gPath("geneMap", "segments"),
##Clear the old graphics object from the display before drawing the modified heat map:
# now the colour of line segments connecting the SNP
# positions to the LD heat map has been changed from black to orange.
#### Draw a resized heat map (in a 'blue-to-red' color scale ####
pushViewport(viewport(width=0.5, height=0.5))
LDheatmap(MyHeatmap, SNP.name = c("rs2283092", "rs6979287"), newpage=FALSE,
#### Draw and modify two heat maps on one plot ####
##Draw and the first heat map on the left half of the graphics device
pushViewport(viewport(x=0, width=0.5, just="left"))
LD1<-LDheatmap(MyHeatmap, color=grey.colors(20), newpage=FALSE,
              title="Pairwise LD in grey.colors(20)",
              SNP.name="rs6979572", geneMapLabelX=0.6,
              geneMapLabelY=0.4, name="ld1")
##Draw the second heat map on the right half of the graphics device
LD2<-LDheatmap(MyHeatmap, newpage=FALSE, title="Pairwise LD in heat.colors(20)",
              SNP.name="rs6979572", geneMapLabelX=0.6, geneMapLabelY=0.4, name="ld2")
##Modify the text size of main title of the first heat map.
grid.edit(gPath("ld1", "heatMap","title"), gp=gpar(cex=1.5))
##Modify the text size and color of the SNP label of the second heat map.
grid.edit(gPath("ld2", "geneMap","SNPnames"), gp=gpar(cex=1.5, col="DarkRed"))
#### Draw a lattice-like plot with heat maps in panels ####
# Make a variable which indicates Chinese vs. Japanese
pop <- factor(c(rep("chinese",45), rep("japanese",45)))
xyplot(1:nrow(CHBJPTSNP) ~ 1:nrow(CHBJPTSNP) | pop,
      type="n", scales=list(draw=FALSE), xlab="", ylab="",
      panel=function(x, y, subscripts,...) {
        LDheatmap(CHBJPTSNP[subscripts,], CHBJPTDist, newpage=FALSE) })
xyplot(1:n ~ 1:n | GIMAP5$subject.support$pop,
      type="n", scales=list(draw=FALSE), xlab="", ylab="",
      panel=function(x, y, subscripts,...) {
                  GIMAP5$snp.support$Position, SNP.name="rs6598", newpage=FALSE) })
#Reset the user's setting for prompting on the graphics output
#to the original value before running these example commands.

# }
Documentation reproduced from package LDheatmap, version 0.99-7, License: GPL-3

Community examples

Looks like there are no examples yet.