R: Heat map

heat_map {opm}

R Documentation

Heat map

Description

A wrapper for heatmap from the stats package and heatmap.2 from the gplots package with some adaptations likely to be useful for OmniLog(R) data. The data-frame and OPMS methods extract a numeric matrix from a given data frame or OPMS object and pass the result to the matrix method.

Usage

  ## S4 method for signature 'MOPMX'
heat_map(object, as.labels,
    subset = opm_opt("curve.param"), as.groups = NULL, sep = " ",
    extract.args = list(), ...) 
  ## S4 method for signature 'OPMS'
heat_map(object, as.labels,
    subset = opm_opt("curve.param"), as.groups = NULL, sep = " ",
    extract.args = list(), ...) 
  ## S4 method for signature 'data.frame'
heat_map(object, as.labels, as.groups = NULL, sep = " ", ...)
  ## S4 method for signature 'matrix'
heat_map(object,
    hclustfun = "ward.D2", distfun = "euclidean", scale = "none",
    r.groups = "row.groups", r.col = opm_opt("colors"),
    c.groups = "col.groups", c.col = opm_opt("colors"),
    magnif = 4, cexRow = magnif[1L] / sqrt(nrow(object)),
    cexCol = magnif[length(magnif)] / sqrt(ncol(object)),
    borders = c(0.55, 0.75),
    margins = if (use.fun[1L] == "gplots")
      c(borders[1L] * cexCol * max(nchar(colnames(object))),
      borders[length(borders)] * cexRow * max(nchar(rownames(object))))
    else
      c(5, 5),
    col = opm_opt("heatmap.colors"), asqr = FALSE, log1 = FALSE, lmap = 1L:3L,
    abbrev = c("none", "row", "column", "both"), plot.na = "plot.NA",
    reorderfun = function(d, w) reorder(d, w, mean), ...,
    use.fun = c("gplots", "stats"))

Arguments

`object`	Matrix, data frame or `OPMS` or `MOPMX` object. The matrix method is mainly designed for curve-parameter matrices as created by `extract` but can be used with any numeric matrix. If a data frame, it must contain at least one column with numeric data. Not all `MOPMX` objects are suitable for this function; see the remarks under `extract`.
`as.labels`	Character, numeric or logical vector indicating the positions of the columns to be joined and used as row labels. If `NULL` or empty, the row names of `object` are used. See `extract` for details.
`as.groups`	Character, numeric or logical vector indicating the positions of the columns to be joined and used as group indicators. If `NULL` or empty, groups are ignored.
`sep`	Character scalar determining how to join row and group names. See `extract` for details.
`subset`	Character scalar passed to the `OPMS` method of `extract`.
`extract.args`	Optional list of arguments passed to that method.
`hclustfun`	Determines the clustering method used. If a function, used directly. If a character scalar, used as the ‘method’ argument of `hclust`. If a list, passed as argument list to `hclust`.
`distfun`	Determines the distance method used. If a function, used directly. If a character scalar, used as the ‘method’ argument of `dist`. If a list, passed as argument list to `dist`.
`scale`	Character scalar. See `heatmap` for details. The default was changed to no rescaling because the curve parameters estimated from OmniLog(R) data have the same scale. If the relative changes per substrate are of interest, ‘column’ should be used.
`r.groups`	Determines the plotting of a colour bar indicating row groups. If `NULL`, ignored. If a function, applied to the row names of `object`; should then yield one group name for each row name. If a character scalar, the name of an attribute of `object` that contains the row group affiliations (ignored if this is not found). Otherwise, coerced to ‘character’ mode. Finally the groups are converted to a factor and used for selecting from `r.col`.
`r.col`	Character vector of colour names used by `r.groups`. Ignored if that is `NULL`.
`c.groups`	Determines the plotting of a colour bar indicating column groups. If `NULL`, ignored. If a function, applied to the column names of `object`; should then yield one group name for each column name. If a character scalar, the name of an attribute of `object` that contains the column group affiliations (ignored if this is not found). Otherwise, coerced to ‘character’ mode. Finally the groups are converted to a factor and used for selecting from `c.col`.
`c.col`	Character vector of colour names used by `c.groups`. Ignored if that is `NULL`.
`magnif`	Numeric vector. Factor(s) used per default by `cexRow` and `cexCol`.
`cexRow`	Magnification of the row labels.
`cexCol`	Magnification of the column labels.
`borders`	Numeric vector. Factor(s) used per default by `margin` and `cexCol`.
`margins`	Two-element numeric vector determining the relative size of the margin (i) at the bottom and (ii) at the left.
`col`	Character vector containing the proper heat map colours.
`asqr`	Logical scalar indicating whether the data should be treated with the arcsine-square root transformation. This usually only makes sense for proportion data and cannot be used in conjunction with the `log1` argument set to `TRUE`. If `NA`, percentages are assumed.
`log1`	Logical scalar indicating whether `log1p` should be used for transforming the data prior to plotting.
`lmap`	Numeric scalar with at least three elements, or empty. If empty, ignored. Otherwise used for mapping logical values to numeric values. See `map_values` for details. Ignored if the data are not logical.
`abbrev`	Character scalar indicating whether row or column shall be abbreviated before plotting. Note that abbreviation is done by shortening words and ending them with a dot, so there is no guarantee that a certain maximum length will be obtained.
`plot.na`	Character scalar with the name of an optional attribute that contains replacement values for `NA` that are inserted prior to plotting.
`reorderfun`	Function passed to `heatmap` or `heatmap.2`. The modification from their defaults is intended to yield an ordering that more strongly reflects row and column sums (instead of also taking cluster size into consideration).
`...`	Optional arguments passed to `heatmap` or `heatmap.2`. Note that some defaults of `heatmap.2` are overwritten even though this is not transparent from the argument list of `heat_map`. If set explicitly, the default `heatmap.2` behaviour is restored. `...` also represents all arguments passed from the `OPMS` or data-frame methods to the matrix method.
`use.fun`	Character scalar. If `gplots`, it is attempted to load the gplots package and use its `heatmap.2` function (the default). If this fails, a warning is issued, and `heatmap` from the stats package (the default) is called instead.

Value

A list as output by heatmap or heatmap.2 with the additional entries rowColMap or colColMap giving the mapping(s) of group names to colours as named character vector(s), if this feature was used.

Examples

# temporarily disabled until an error specific for Windows on
# R-Forge is fixed; needs not affect user code, Windows users
# can well try heat_map()
if (!grepl("windows", Sys.info()[["sysname"]], TRUE, TRUE)) {

# Matrix method (usually unnecessary, see below)
x <- extract(vaas_4, as.labels = list("Strain"),
  as.groups = list("Species"))
hm <- heat_map(x)
stopifnot(identical(metadata(vaas_4, "Species"), names(hm$rowColMap)))

# 'OPMS' method (more convenient)
hm.2 <- heat_map(vaas_4, as.labels = "Strain", as.groups = "Species")
stopifnot(identical(hm[-3], hm.2[-3]))

# Data-frame method
x <- extract(vaas_4, as.labels = list("Species", "Strain"), dataframe = TRUE)
hm <- heat_map(x, as.labels = "Strain", as.groups = "Species")
stopifnot(identical(metadata(vaas_4, "Species"), names(hm$rowColMap)))

}

plot of chunk unnamed-chunk-1