heatmapPlot: displays the heatmap based on the data from heatmapData
In compEpiTools: Tools for computational epigenomics

Description Usage Arguments Details Value References See Also Examples

displays heatmap of counts for a list of GRanges, typically computed based on the heatmapData function

1
2
3

heatmapPlot(matList, sigMat=NULL, qnorm=NULL, tnorm=NULL,
rowLab=FALSE, colLab=TRUE, margins=NULL, colors=NULL,
clusterInds=1:length(matList), dendrogram=TRUE)

`matList`	list; a list of matrices, all with the same number of rows and columns, typically the output of heatmapData
`sigMat`	matrix; a matrix of p-values with nrow equal to the nrow of the matList matrices and ncol equal to the length of matList
`qnorm`	array; an array with length equal to the length of matList containing either NULL or thresholds for quantile normalization, see details
`tnorm`	array; an aray with length equal to the length of matList containing either NULL or threshold for data normalization, see details
`rowLab`	logical; whether to add row labels to the heatmap, taken from the rownames of matList[[1]]
`colLab`	logical; whether to add column names to the heatmap, taken from the names of matList
`colors`	either NULL or character with an array of valid colors; it only works if sigMat is NULL
`margins`	either NULL or a numeric array of length 2
`clusterInds`	either NULL or numerics, defining with matList items have to be used to drive the clustering of heatmap rows
`dendrogram`	logical; whether to display the dendrogram folling the clustering of rows

Each matrix in matList is either ranging from 0 to 1 or will be forced to by dividing to its maxumum. Alternatively, matrix normalization can be obtained using qnorm or tnorm. Setting qrnorm to X [0,1] for a given matList matrix will force the maximum of the matrix to be quantile(matrix, X). Setting trnorm to X for a given matList matrix will force the maximum of the matrix to be X. Using either qnorm or tnorm matrix will be finally normalized to 1 by dividing it by its maximum.

If sigMat is not NULL it is expected to contain a list of pvalues or scores [0,1] for each range for each matList dataset. The colorscale of the heatmap will be adjusted to display the significance, by lightening the observation colors as a function of its significance, see the example. A minimum pvalue of 1e-10 is forced. 50 levels of intensity (white to orange to red palette, as displayed in the colorscale) and 10 levels of significance (white for the less significance, full color according to the intensity palette, as displayed in the colorsclae) are considered. Plese ignore the values reported in the colorscale. If sigMat is NULL, the normalized intensity in each matList item is reported as it is on a white to beige to red default palette, or based on the colors defined in the colors argument.

If margins is set to NULL the row and columns margins used to display labels are computed automatically, otherwise a numeric array of length two can be set to define them (in lines).

clusterInds can be used to define which matList items drive the clustering of heatmap rows. If clusterInds is NULL no clustering is performed and no dendrogram is displayed. If clusterInds is an array of index in 1:length(matList), only those matList items will be used to determine the clustering and the dendrogram, while all matList data will be displayed.

If a TxDB was provided to heatmapData beofre calling heatmapPlot, the last two tracks are about the overlap with exons and introns in the forward and reverse strand, repectively. If the default white to red color palette is used, and sigMat is NULL exons will be plotted in red and introns in pink. Rather, if sigMat is defined, introns will be in orange.

A list with two items

`data`	the normalized matrix used for the final heatmap visualization
`heatRes`	the list invisibly returned by the heatmap.2 function, see its documentation

Be careful that the rowInds contained in heatRes poiting to the new order of clustered data rows, is intended to list the reordered rows starting from the bottom of the heatmap.

http://genomics.iit.it/groups/computational-epigenomics.html

See Also as heatmap.2

require(TxDb.Mmusculus.UCSC.mm9.knownGene)
txdb <- TxDb.Mmusculus.UCSC.mm9.knownGene
isActiveSeq(txdb) <- c(TRUE, rep(FALSE, length(isActiveSeq(txdb)) - 1))
TSSpos <- TSS(txdb)
names(TSSpos) <- TSSpos$tx_name
gr <- TSSpos[1:50]
start(gr) <- start(gr) - 1000
end(gr) <- end(gr) - 600
pvalues <- c(runif(20,1e-20,1e-8), runif(15,1e-4,1e-2), runif(15,0.5,1))
mcols(gr) <- pvalues
extgr <- GRanges(seqnames(gr), ranges= IRanges(start(gr) - 1000, end(gr) + 1000))
data <- heatmapData(grl=list(ChIPseq=gr), refgr=extgr, type='gr', useScore=TRUE,
  Nnorm=TRUE, Snorm=TRUE, nbins=6, txdb=txdb)
rownames(data[[1]][[1]]) <- paste(1:50, signif(pvalues,1), sep=' # ')
heatmapPlot(matList=data$matList, qnorm=NULL, tnorm=NULL, 
  rowLab=TRUE, colLab=TRUE, clusterInds=1:3)
dev.new()
heatmapPlot(matList=data$matList, sigMat=data$scoreMat, qnorm=NULL, tnorm=NULL, 
  rowLab=TRUE, colLab=TRUE, clusterInds=1:3)
restoreSeqlevels(txdb)