KolmogorovMinDist {RobLoxBioC}

Generic Function for Computing Minimum Kolmogorov Distance for Biological Data

Description

Generic function for computing minimum Kolmogorov distance for biological data.

Usage

KolmogorovMinDist(x, D, ...)

## S4 method for signature 'matrix,Norm'
KolmogorovMinDist(x, D, mad0 = 1e-4)

## S4 method for signature 'AffyBatch,AbscontDistribution'
KolmogorovMinDist(x, D, bg.correct = TRUE, pmcorrect = TRUE, 
                  verbose = TRUE)

## S4 method for signature 'beadLevelData,AbscontDistribution'
KolmogorovMinDist(x, D, log = FALSE, what = "Grn", 
                  probes = NULL, arrays = NULL)

Arguments

x

biological data.

D

object of class "UnivariateDistribution".

...

additional parameters.

mad0

scale estimate used if computed MAD is equal to zero. Median and MAD are used as start parameter for optimization.

bg.correct

if TRUE MAS 5.0 background correction is performed; confer bg.correct.mas.

pmcorrect

if TRUE log2(PM/MM) is used. If FALSE only log2(PM) is used.

verbose

logical: if TRUE, some messages are printed.

log

if TRUE, then the log2 intensities for each bead-type are summarized.

what

character string specifying which intensities/values to summarize; see getBeadData.

probes

Specify particular probes to summarize. If left NULL then all the probes on the first array are used.

arrays

integer (scalar or vector) specifying the strips/arrays to summarize. If NULL, then all strips/arrays are summarized.

Details

The minimum Kolmogorov distance is computed for each row of a matrix, each Affymetrix probe, or each Illumina bead, respectively.

So far, only the minimum distance to the set of normal distributions can be computed.

Value

List with components dist containing a numeric vector or matrix with minimum Kolmogorov distances and n a numeric vector or matrix with the corresponding sample sizes.

Author(s)

Matthias Kohl [e-mail]

References

Huber, P.J. (1981) Robust Statistics. New York: Wiley.

Rieder, H. (1994) Robust Asymptotic Statistics. New York: Springer.

See Also

KolmogorovDist, MDEstimator

Examples

set.seed(123) # to have reproducible results for package checking

## matrix method for KolmogorovMinDist
ind <- rbinom(200, size=1, prob=0.05) 
X <- matrix(rnorm(200, mean=ind*3, sd=(1-ind) + ind*9), nrow = 2)
KolmogorovMinDist(X, D = Norm())

## using Affymetrix data
data(SpikeIn)
probes <- log2(pm(SpikeIn))
(res <- KolmogorovMinDist(probes, Norm()))
boxplot(res$dist)


## \donttest because of check time
## using Affymetrix data
library(affydata)
data(Dilution)
res <- KolmogorovMinDist(Dilution[,1], Norm())
summary(res$dist)
boxplot(res$dist)
plot(res$n, res$dist, pch = 20, main = "Kolmogorov distance vs. sample size",
     xlab = "sample size", ylab = "Kolmogorov distance",
     ylim = c(0, max(res$dist)))
uni.n <- min(res$n):max(res$n)
lines(uni.n, 1/(2*uni.n), col = "orange", lwd = 2)
legend("topright", legend = "minimal possible distance", fill = "orange")

## Illumina bead level data
library(beadarrayExampleData)
data(exampleBLData)
res <- KolmogorovMinDist(exampleBLData, Norm(), arrays = 1)
res1 <- KolmogorovMinDist(exampleBLData, Norm(), log = TRUE, arrays = 1)
summary(cbind(res$dist, res1$dist))
boxplot(list(res$dist, res1$dist), names = c("raw", "log-raw"))
sort(unique(res1$n))
plot(res1$n, res1$dist, pch = 20, main = "Kolmogorov distance vs. sample size",
     xlab = "sample size", ylab = "Kolmogorov distance",
     ylim = c(0, max(res1$dist)), xlim = c(min(res1$n), 56))
uni.n <- min(res1$n):56
lines(uni.n, 1/(2*uni.n), col = "orange", lwd = 2)
legend("topright", legend = "minimal possible distance", fill = "orange")