R: Linear model-based testing for summarized imaging experiments

meansTest-methods {Cardinal}

R Documentation

Linear model-based testing for summarized imaging experiments

Description

Performs hypothesis testing for imaging experiments by fitting linear mixed models to summarizations or segmentations.

Usage

## S4 method for signature 'SparseImagingExperiment'
meansTest(x, fixed, random, groups = run(x),
        BPPARAM = getCardinalBPPARAM(), ...)

## S4 method for signature 'SparseImagingExperiment'
segmentationTest(x, fixed, random, groups = run(x),
        classControl = c("Ymax", "Mscore"),
        BPPARAM = getCardinalBPPARAM(), ...)

## S4 method for signature 'SpatialDGMM'
segmentationTest(x, fixed, random, model = modelData(x),
        classControl = c("Ymax", "Mscore"),
        BPPARAM = getCardinalBPPARAM(), ...)

## S4 method for signature 'MeansTest'
summary(object, ..., BPPARAM = getCardinalBPPARAM())

## S4 method for signature 'SegmentationTest'
summary(object, ..., BPPARAM = getCardinalBPPARAM())

Arguments

`x`	An imaging dataset or segmented/summarized imaging dataset.
`fixed`	A one-sided formula giving the fixed effects of the model on the RHS. The response will added to the LHS, and the formula will be passed to the underlying modeling function.
`random`	A one-sided formula giving the random effects of the model on the RHS. See `lme` for the allowed specifications.
`groups`	The summarization units. Pixels from different groups will be segmented/summarized separately. Each distinct observational unit (e.g., tissue sample) should be assigned to a unique group.
`model`	An integer vector or `list` specifying which fitted model to plot. If this is an integer vector, it should give the rows indices of `modelData(x)` to use for plotting. Otherwise, it should be a list giving the values of parameters in `modelData(x)`.
`classControl`	Either the method used to match segmented classes to the fixed effects, or a list where each element is a vector of name-value pairs giving the mapping between groups and classes (e.g., c(group1=class1, group2=class2, ...)). For automated matching methods, 'Ymax' means to use the classes with the highest mean response for each group, and 'Mscore' means to select classses based on a match score quantifying the overlap between classes and fixed effects.
`...`	Passed to internal linear modeling methods.
`object`	A fitted model object to summarize.
`BPPARAM`	An optional instance of `BiocParallelParam`. See documentation for `bplapply`.

Value

An object of class MeansTest or SegmentationTest, which is a ImagingResult, where each element of the resultData slot contains at least the following components:

model:: A linear model fitted using either lm or lme.
data:: The summarized data used to fit the model.

Author(s)

Dan Guo and Kylie A. Bemis

Examples

set.seed(1)
x <- simulateImage(preset=4, nruns=3, npeaks=10,
    dim=c(10,10), peakheight=5, peakdiff=2,
    representation="centroid")

groups <- replace(run(x), !(x$circleA | x$circleB), NA)

fit <- meansTest(x, ~ condition, groups=groups)

summary(fit)

[Package Cardinal version 2.6.0 Index]