DataTrack-class: DataTrack class and methods
In ivanek/Gviz: Plotting data and annotation information along genomic coordinates

DataTrack-class

R Documentation

DataTrack class and methods

Description

A class to store numeric data values along genomic coordinates. Multiple samples as well as sample groupings are supported, with the restriction of equal genomic coordinates for a single observation across samples.

Usage

## S4 method for signature 'DataTrack'
initialize(.Object, data = matrix(), strand, ...)

## S4 method for signature 'ReferenceDataTrack'
initialize(
  .Object,
  stream,
  reference,
  mapping = list(),
  args = list(),
  defaults = list(),
  ...
)

DataTrack(
  range = NULL,
  start = NULL,
  end = NULL,
  width = NULL,
  data,
  chromosome,
  strand,
  genome,
  name = "DataTrack",
  importFunction,
  stream = FALSE,
  ...
)

## S4 method for signature 'DataTrack'
values(x, all = FALSE)

## S4 replacement method for signature 'DataTrack'
values(x) <- value

## S4 method for signature 'DataTrack'
strand(x)

## S4 replacement method for signature 'DataTrack,ANY'
strand(x) <- value

## S4 method for signature 'DataTrack,ANY'
split(x, f, drop = FALSE, ...)

## S4 method for signature 'DataTrack'
feature(GdObject)

## S4 replacement method for signature 'DataTrack,character'
feature(GdObject) <- value

## S4 method for signature 'DataTrack'
collapseTrack(GdObject, diff = .pxResolution(coord = "x"), xrange)

## S4 method for signature 'DataTrack,ANY,ANY,ANY'
x[i, j, ..., drop = FALSE]

## S4 method for signature 'DataTrack'
subset(
  x,
  from = NULL,
  to = NULL,
  sort = FALSE,
  drop = TRUE,
  use.defaults = TRUE,
  ...
)

## S4 method for signature 'ReferenceDataTrack'
subset(x, from, to, chromosome, ...)

## S4 method for signature 'DataTrack'
drawAxis(GdObject, ...)

## S4 method for signature 'DataTrack'
drawGD(GdObject, minBase, maxBase, prepare = FALSE, subset = TRUE, ...)

## S4 method for signature 'DataTrack'
show(object)

## S4 method for signature 'ReferenceDataTrack'
show(object)

Arguments

`.Object`	.Object
`data`	A numeric matrix of data points with the number of columns equal to the number of coordinates in `range`, or a numeric vector of appropriate length that will be coerced into such a one-row matrix. Each individual row is supposed to contain data for a given sample, where the coordinates for each single observation are constant across samples. Depending on the plotting type of the data (see 'Details' and 'Display Parameters' sections), sample grouping or data aggregation may be available. Alternatively, this can be a character vector of column names that point into the element metadata of the `range` object for subsetting. Naturally, this is only supported when the `range` argument is of class `GRanges`.
`strand`	Character vector, the strand information for the individual track items. Currently this has to be unique for the whole track and doesn't really have any visible consequences, but we might decide to make `DataTracks` strand-specific at a later stage.
`...`	Additional items which will all be interpreted as further display parameters.
`stream`	A logical flag indicating that the user-provided import function can deal with indexed files and knows how to process the additional `selection` argument when accessing the data on disk. This causes the constructor to return a `ReferenceDataTrack` object which will grab the necessary data on the fly during each plotting operation.
`range`	An optional meta argument to handle the different input types. If the `range` argument is missing, all the relevant information to create the object has to be provided as individual function arguments (see below). The different input options for `range` are: A `GRanges` object: essentially all the necessary information to create a `DataTrack` can be contained in a single `GRanges` object. The track's coordinates are taken from the `start`, `end` and `seqnames` slots, the genome information from the genome slot, and the numeric data values can be extracted from additional metadata columns columns (please note that non-numeric columns are being ignored with a warning). As a matter of fact, calling the constructor on a `GRanges` object without further arguments, e.g. `DataTrack(range=obj)` is equivalent to calling the coerce method `as(obj, "DataTrack")`. Alternatively, the `GRanges` object may only contain the coordinate information, in which case the numeric data part is expected to be present in the separate `data` argument, and the ranges have to match the dimensions of the data matrix. If `data` is not `NULL`, this will always take precedence over anything defined in the `range` argument. See below for details. An `IRanges` object: this is very similar to the above case, except that the numeric data part now always has to be provided in the separate `data` argument. Also the chromosome information must be provided in the `chromosome` argument, because neither of the two can be directly encoded in an `IRange` object. A `data.frame` object: the `data.frame` needs to contain at least the two mandatory columns `start` and `end` with the range coordinates. It may also contain a `chromosome` column with the chromosome information for each range. If missing it will be drawn from the separate `chromosome` argument. All additional numeric columns will be interpreted as data columns, unless the `data` argument is explicitely provided. A `character` scalar: in this case the value of the `range` argument is considered to be a file path to an annotation file on disk. A range of file types are supported by the `Gviz` package as identified by the file extension. See the `importFunction` documentation below for further details.
`start, end, width`	Integer vectors, giving the start and the end end coordinates for the individual track items, or their width. Two of the three need to be specified, and have to be of equal length or of length one, in which case the single value will be recycled accordingly. Otherwise, the usual R recycling rules for vectors do not apply and the function will cast an error.
`chromosome`	The chromosome on which the track's genomic ranges are defined. A valid UCSC chromosome identifier if `options(ucscChromosomeNames=TRUE)`. Please note that in this case only syntactic checking takes place, i.e., the argument value needs to be an integer, numeric character or a character of the form `chrx`, where `x` may be any possible string. The user has to make sure that the respective chromosome is indeed defined for the the track's genome. If not provided here, the constructor will try to construct the chromosome information based on the available inputs, and as a last resort will fall back to the value `chrNA`. Please note that by definition all objects in the `Gviz` package can only have a single active chromosome at a time (although internally the information for more than one chromosome may be present), and the user has to call the `chromosome<-` replacement method in order to change to a different active chromosome.
`genome`	The genome on which the track's ranges are defined. Usually this is a valid UCSC genome identifier, however this is not being formally checked at this point. If not provided here the constructor will try to extract this information from the provided input, and eventually will fall back to the default value of `NA`.
`name`	Character scalar of the track's name used in the title panel when plotting.
`importFunction`	A user-defined function to be used to import the data from a file. This only applies when the `range` argument is a character string with the path to the input data file. The function needs to accept an argument `file` containing the file path and has to return a proper `GRanges` object with the data part attached as numeric metadata columns. Essentially the process is equivalent to constructing a `DataTrack` directly from a `GRanges` object in that non-numeric columns will be dropped, and further subsetting can be archived by means of the `data` argument. A set of default import functions is already implemented in the package for a number of different file types, and one of these defaults will be picked automatically based on the extension of the input file name. If the extension can not be mapped to any of the existing import function, an error is raised asking for a user-defined import function. Currently the following file types can be imported with the default functions: `wig`, `bigWig/bw`, `bedGraph` and `bam`. Some file types support indexing by genomic coordinates (e.g., `bigWig` and `bam`), and it makes sense to only load the part of the file that is needed for plotting. To this end, the `Gviz` package defines the derived `ReferenceDataTrack` class, which supports streaming data from the file system. The user typically does not have to deal with this distinction but may rely on the constructor function to make the right choice as long as the default import functions are used. However, once a user-defined import function has been provided and if this function adds support for indexed files, you will have to make the constructor aware of this fact by setting the `stream` argument to `TRUE`. Please note that in this case the import function needs to accept a second mandatory argument `selection` which is a `GRanges` object containing the dimensions of the plotted genomic range. As before, the function has to return an appropriate `GRanges` object.
`value`	Value to be set.
`GdObject`	Object of `GdObject-class`.

Details

Depending on the setting of the type display parameter, the data can be plotted in various different forms as well as combinations thereof. Supported plotting types are:

p:: simple xy-plot.
l:: lines plot. In the case of multiple samples this plotting type is not overly usefull since the points in the data matrix are connected in column-wise order. Type a might be more appropriate in these situations.
b:: combination of xy-plot and lines plot.
a:: lines plot of the column-wise average values.
s:: sort and connect data points along the x-axis
S:: sort and connect data points along the y-axis
g:: add grid lines. To ensure a consitant look and feel across multiple tracks, grid lines should preferentially be added by using the grid display parameter.
r:: add a regression line to the plot.
h:: histogram-like vertical lines centered in the middle of the coordinate ranges.
smooth:: add a loess fit to the plot. The following display parameters can be used to control the loess calculation: span, degree, family, evaluation. See panel.loess for details.
histogram:: plot data as a histogram, where the width of the histogram bars reflects the width of the genomic ranges in the range slot.
mountain:: plot a smoothed version of the data relative to a baseline, as defined by the baseline display parameter. The following display parameters can be used to control the smoothing: span, degree, family, evaluation. See panel.loess for details. The layout of the plot can be further customized via the following display parameters: col.mountain, lwd.mountain, lty.mountain, fill.mountain.
polygon:: plot data as a polygon (similar to mountain-type but without smoothing). Data are plotted relative to a baseline, as defined by the baseline display parameter. The layout of the plot can be further customized via the following display parameters: col.mountain, lwd.mountain, lty.mountain, fill.mountain.
boxplot:: plot the data as box-and-whisker plots. The layout of the plot can be further customized via the following display parameters: box.ratio, box.width, varwidt, notch, notch.frac, levels.fos, stats, coef, do.out. See panel.bwplot for details.
gradient:: collapse the data across samples and plot this average value as a color-coded gradient. Essenitally this is similar to the heatmap-type plot of a single sample. The layout of the plot can be further customized via the display parameters ncolor and gradient which control the number of gradient colors as well as the gradient base colors, respectively.
heatmap:: plot the color-coded values for all samples in the form of a heatmap. The data for individual samples can be visually separated by setting the separator display parameter. It's value is taken as the amount of spacing in pixels in between two heatmap rows. The layout of the plot can be further customized via the display parameters ncolor and gradient which control the number of gradient colors as well as the gradient base colors, respectively.
horizon:: plot continuous data by cutting the y range into segments and overplotting them with color representing the magnitude and direction of deviation. This is particularly useful when comparing multiple samples, in which case the horizon strips are stacked. See horizonplot for details. Please note that the origin and horizonscale arguments of the Lattice horizonplot function are available as display parameters horizon.origin and horizon.scale.

For some of the above plotting-types the groups display parameter can be used to indicate sample sub-groupings. Its value is supposed to be a factor vector of similar length as the number of samples. In most cases, the groups are shown in different plotting colors and data aggregation operations are done in a stratified fashion.

The window display parameter can be used to aggregate the data prior to plotting. Its value is taken as the number of equal-sized windows along the genomic coordinates of the track for which to compute average values. The special value auto can be used to automatically determine a reasonable number of windows which can be particularly useful when plotting very large genomic regions with many data points.

The aggregation parameter can be set to define the aggregation function to be used when averaging in windows or across collapsed items. It takes the form of either a function which should condense a numeric vector into a single number, or one of the predefined options as character scalars "mean", "median" or "sum" for mean, median or summation, respectively. Defaults to computing mean values for each sample. Note that the predefined options can be much faster because they are optimized to work on large numeric tables.

Value

The return value of the constructor function is a new object of class DataTrack or ReferenceDataTrack.

Functions

initialize(DataTrack): Initialize.
ReferenceDataTrack-class: The file-based version of the DataTrack-class.
initialize(ReferenceDataTrack): Initialize.
DataTrack(): Constructor function for DataTrack-class
values(DataTrack): return the raw data values of the object, i.e., the data matrix in the data slot.
values(DataTrack) <- value: Replace the data matrix in the data slot.
strand(DataTrack): return a vector of strand specifiers for all track items, in the form '+' for the Watson strand, '-' for the Crick strand or '*' for either of the two.
strand(x = DataTrack) <- value: replace the strand information for the track items. The replacement value needs to be an appropriate scalar or vector of strand values.
split(x = DataTrack, f = ANY): Split a DataTrack object by an appropriate factor vector (or another vector that can be coerced into one). The output of this operation is a list of DataTrack objects.
feature(DataTrack): returns NULL since there is no grouping information for the ranges in a DataTrack.
feature(GdObject = DataTrack) <- value: this return the unaltered input object since there is no grouping information for the ranges in a DataTrack.
collapseTrack(DataTrack): preprocess the track before plotting. This will collapse overlapping track items based on the available resolution and increase the width and height of all track objects to a minimum value to avoid rendering issues. See collapsing for details.
x[i: subset the items in the DataTrack object. This is essentially similar to subsetting of the GRanges object in the range slot. For most applications, the subset method may be more appropriate.
subset(DataTrack): Subset a DataTrack by coordinates and sort if necessary.
subset(ReferenceDataTrack): Subset a ReferenceDataTrack by coordinates and sort if necessary.
drawAxis(DataTrack): add a y-axis to the title panel of a track.
drawGD(DataTrack): plot the object to a graphics device. The return value of this method is the input object, potentially updated during the plotting operation. Internally, there are two modes in which the method can be called. Either in 'prepare' mode, in which case no plotting is done but the object is preprocessed based on the available space, or in 'plotting' mode, in which case the actual graphical output is created. Since subsetting of the object can be potentially costly, this can be switched off in case subsetting has already been performed before or is not necessary.
show(DataTrack): Show method.
show(ReferenceDataTrack): Show method.

Objects from the class

Objects can be created using the constructor function DataTrack.

Author(s)

Florian Hahne

Examples

## Object construction:

## An empty object
DataTrack()

## from individual arguments
dat <- matrix(runif(400), nrow = 4)
dtTrack <- DataTrack(
    start = seq(1, 1000, len = 100), width = 10, data = dat,
    chromosome = 1, genome = "mm9", name = "random data"
)

## from GRanges
library(GenomicRanges)
gr <- GRanges(seqnames = "chr1", ranges = IRanges(seq(1, 1000, len = 100),
    width = 10
))
values(gr) <- t(dat)
dtTrack <- DataTrack(range = gr, genome = "mm9", name = "random data")

## from IRanges
dtTrack <- DataTrack(
    range = ranges(gr), data = dat, genome = "mm9",
    name = "random data", chromosome = 1
)

## from a data.frame
df <- as.data.frame(gr)
colnames(df)[1] <- "chromosome"
dtTrack <- DataTrack(range = df, genome = "mm9", name = "random data")


## Plotting
plotTracks(dtTrack)

## Track names
names(dtTrack)
names(dtTrack) <- "foo"
plotTracks(dtTrack)

## Subsetting and splitting
subTrack <- subset(dtTrack, from = 100, to = 300)
length(subTrack)
subTrack[1:2, ]
subTrack[, 1:2]
split(dtTrack, rep(1:2, each = 50))

## Accessors
start(dtTrack)
end(dtTrack)
width(dtTrack)
position(dtTrack)
width(subTrack) <- width(subTrack) - 5

strand(dtTrack)
strand(subTrack) <- "-"

chromosome(dtTrack)
chromosome(subTrack) <- "chrX"

genome(dtTrack)
genome(subTrack) <- "mm9"

range(dtTrack)
ranges(dtTrack)

## Data
values(dtTrack)
score(dtTrack)

## coercion
as(dtTrack, "data.frame")

ivanek/Gviz documentation built on Nov. 20, 2023, 8:16 p.m.