kernelize: Smooth expression values or scores
In macat: MicroArray Chromosome Analysis Tool

Description Usage Arguments Value Author(s) See Also Examples

'kernelize' uses a kernel to smooth the data given in geneLocations by computing a weighted sum of the values vector. The weights for each position are given in the kernelweights matrix. A kernelweights matrix can be obtained by using the kernelmatrix function.

1
2
3

getsteps(geneLocations, step.width)
kernelmatrix(steps, geneLocations, kernel, kernelparams)
kernelize(values, kernelweights)

`geneLocations`	a list of gene locations (length n)
`step.width`	the width of steps in basepairs
`steps`	a list of locations where the kernelization shall be computed
`kernel`	kernel function one of rbf, kNN or basePairDistance (or your own)
`kernelparams`	a list of named parameters for the kernel (default is fitted to the data)
`values`	vector of length n or matrix (m x n) of values that are to be smoothed
`kernelweights`	a matrix of (n x steps) where n is the length of the values vector and steps is the number of points where you wish to interpolate

`getsteps`	a list of locations starting at min(genLocations) going to max(geneLocations) with steps of size step.width
`kernelmatrix`	a matrix of (n x steps) containing the kernel weights for each location in steps
`kernelize`	a vector of length steps or a matrix (m x steps) containing the smoothed values

MACAT Development team

compute.sliding, evalScoring

  data(stjd)
  genes = seq(100)
  geneLocations = abs(stjd$geneLocation[genes])
  geneExpression = stjd$expr[genes,]
  step.width = 100000
  steps = getsteps(geneLocations, step.width)
  weights = kernelmatrix(steps, geneLocations, rbf, list(gamma=1/10^13))
  kernelized = kernelize(geneExpression, weights)
  plot(steps, kernelized[1,])