CSRPeaksFilter {SPUTNIK}

Performs the peak selection based on complete spatial randomness test.

Description

CSRPeaksFilter returns the significance for the null hypothesis that the spatial distribution of the peak intensities follow a random pattern. A significant p-value (q-values can be returned after applying multiple testing correction) allows to reject the hypothesis that the spatial distribution of a peak signal is random. The tests are performed using the functions available in the statspat R package.

Usage

CSRPeaksFilter(
  msiData,
  method = "ClarkEvans",
  covariateImage = NULL,
  adjMethod = "bonferroni",
  returnQvalues = TRUE,
  plotCovariate = FALSE,
  cores = 1,
  verbose = TRUE,
  ...
)

Arguments

msiData

msi.dataset-class object. See msiDataset.

method

string (default = "ClarkEvans"). CSR statistical test applied to the peaks signal. Accepted values are:

• "ClarkEvans": performs a test based on the Clark and Evans aggregation R index. This test evaluates the compares of the nearest-neighbors distances to the case of purely random pattern.

• "KS": performs a test of goodness-of-fit between the signal pixels associated point process pattern and a spatial covariate using the Kolmogorov-Smirnov test. The covariate is defined by the reference image.

covariateImage

ms.image-class object. An image used as covariate (required for Kolmogorov-Smirnov test).

adjMethod

string (default = "bonferroni"). Multiple testing correction method. Possible values coincide with those of the stats::p.adjust function.

returnQvalues

logical (default = TRUE). Whether the computed q-values should be returned together with the p-values.

plotCovariate

logical (default = FALSE). Whether the covariate image should be visualized. Read only when method = "KS".

cores

integer (default = 1). Number of CPU cores. Parallel computation if greater than 1.

verbose

logical (default = TRUE). Additional output texts are generated.

...

additional parameters compatible with the statspat.core functions. See cdf.test for "KS" and clarkevans.test. for "ClarkEvans"

Value

List of the p-values and adjusted p-values for the CSR test.

Author(s)

Paolo Inglese [e-mail]

References

Baddeley, A., & Turner, R. (2005). Spatstat: an R package for analyzing spatial point patterns. Journal of statistical software, 12(6), 1-42.

Clark, P.J. and Evans, F.C. (1954) Distance to nearest neighbour as a measure of spatial relationships in populations. Ecology 35, 445–453.

Berman, M. (1986) Testing for spatial association between a point process and another stochastic process. Applied Statistics 35, 54–62.

Examples

## Load package
library("SPUTNIK")

## Mass spectrometry intensity matrix
X <- matrix(rnorm(16000), 400, 40)
X[X < 0] <- 0

## Print original dimensions
print(dim(X))

## m/z vector
mzVector <- seq(600, 900, by = (900 - 600) / 39)

## Read the image size
imSize <- c(20, 20)

## Construct the ms.dataset object
msiX <- msiDataset(X, mzVector, imSize[1], imSize[2])

## Calculate the p-values using the Clark Evans test, then apply Benjamini-
## Hochberg correction.
csr <- CSRPeaksFilter(
  msiData = msiX, method = "ClarkEvans",
  calculateCovariate = FALSE, adjMethod = "BH"
)

## Print selected peaks
print(csr$q.value)

## Create a new filter selecting corrected p-values < 0.001
selIdx <- which(csr$q.value < 0.001)
csrFilter <- createPeaksFilter(selIdx)