SIMageSet: Class to contain pixel-sparse imaging data

Description

An iSet derived class for pixel-sparse imaging data. Data is stored to be memory efficient when there are missing pixels or when the the stored images are non-rectangular regions. The data structures remain efficient for non-sparse pixel data as long as the full datacube does not need to be reconstructed often, and single images or feature vectors are of primary interest. This class can be combined with Hashmat to be sparse in both feature space and pixel space. This is useful for datasets with sparse signals, such as processed spectra.

MSImageSet is a derived class of SImageSet for storing mass spectrometry imaging experiments.

Usage

## Instance creation
SImageSet(
    data = Hashmat(nrow=0, ncol=0),
    coord = expand.grid(
        x = seq_len(prod(dim(data)[-1])),
        y = seq_len(ifelse(prod(dim(data)[-1]) > 0, 1, 0))),
    imageData = SImageData(
        data=data,
        coord=coord),
    pixelData = IAnnotatedDataFrame(
        data=coord,
        varMetadata=data.frame(labelType=rep("dim", ncol(coord)))),
    featureData = AnnotatedDataFrame(
        data=data.frame(row.names=seq_len(nrow(data)))),
    protocolData = AnnotatedDataFrame(
        data=data.frame(row.names=sampleNames(pixelData))),
    experimentData = new("MIAPE-Imaging"),
    ...)

## Additional methods documented below

Arguments

Slots

imageData:

An instance of SImageData, which stores one or more matrices of equal number of dimensions as elements in an 'immutableEnvironment'. This slot preserves copy-on-write behavior when it is modified specifically, but is pass-by-reference otherwise, for memory efficiency.

pixelData:

Contains pixel information in an IAnnotatedDataFrame. This includes both pixel coordinates and phenotypic and sample data. Its rows correspond to the columns in imageData.

featureData:

Contains variables describing features in an IAnnotatedDataFrame. Its rows correspond to the rows in imageData.

experimentData:

Contains details of experimental methods. Should be an object of a derived class of MIAxE.

protocolData:

Contains variables in an IAnnotatedDataFrame describing the generation of the samples in pixelData.

.__classVersion__:

A Versions object describing the version of the class used to created the instance. Intended for developer use.

Extends

iSet, directly. VersionedBiobase, by class "iSet", distance 1. Versioned, by class "VersionedBiobase", distance 2.

Creating Objects

SImageSet instances are usually created through SImageSet().

Methods

Class-specific methods:

iData(object), iData(object) <- value:

Access and set the sparse image data in imageData. This is a matrix-like object with rows corresponding to features and columns corresponding to pixels, so that each column of the returned object is a feature vector.

regeneratePositions:

Regenerates the positionArray in imageData used to reconstruct the datacube based on the coordinates in pixelData. Normally, this should not be called by the user. However, if the coordinates are modified manually, it can be used to re-sync the data structures.

Standard generic methods:

combine(x, y, ...):

Combine two or more SImageSet objects. Unique 'sample's in pixelData are treated as a dimension.

SImageSet[i, j, ..., drop]:

Subset an SImageSet based on the rows (featureData components) and the columns (pixelData components). The result is a new SImageSet.

See iSet for additional methods.

Author(s)

Kylie A. Bemis

See Also

iSet, SImageData, MSImageSet

Examples

## Create an SImageSet object
data <- matrix(1:27, nrow=3)
coord <- expand.grid(x=1:3, y=1:3)
sset <- SImageSet(data=data, coord=coord)

## Access a single image corresponding to the first feature
imageData(sset)[1,,]

## Reconstruct the datacube
imageData(sset)[]

## Access the P x N matrix of column-wise feature vectors
iData(sset)

## Subset the SImageSet to the first 2 features and first 6 pixels
sset2 <- sset[1:2, 1:6]
imageData(sset2)[]
sset2