peakDetection: Detect peaks (local maximum) from values series
In nucleR: Nucleosome positioning package for R

Description Usage Arguments Details Value Note Author(s) See Also Examples

This function allows a efficient recognition of the local maximums (peaks) in a given numeric vector.

peakDetection(
  data,
  threshold = 0.25,
  chromosome = NULL,
  width = 1,
  score = TRUE,
  min.cov = 2,
  mc.cores = 1
)

## S4 method for signature 'list'
peakDetection(
  data,
  threshold = "25%",
  width = 1,
  score = TRUE,
  min.cov = 2,
  mc.cores = 1
)

## S4 method for signature 'numeric'
peakDetection(
  data,
  threshold = "25%",
  chromosome = NULL,
  width = 1,
  score = TRUE,
  min.cov = 2,
  mc.cores = 1
)

`data`	Input numeric values, or a list of them
`threshold`	Threshold value from which the peaks will be selected. Can be given as a percentage string (i.e., `"25\\%"` will use the value in the 1st quantile of `data`) or as an absolute coverage numeric value (i.e., `20` will not look for peaks in regions without less than 20 reads (or reads per milion)).
`chromosome`	Optionally specify the name of the chromosome for input data that doesn't specify it.
`width`	If a positive integer > 1 is given, the peaks are returned as a range of the given width centered in the local maximum. Useful for nucleosome calling from a coverage peak in the dyad.
`score`	If TRUE, the results will be scored using `peakScoring()` function.
`min.cov`	Minimum coverage that a peak needs in order to be considered as a nucleosome call.
`mc.cores`	The number of cores to use, i.e. at most how many child processes will be run simultaneously. Parallelization requires at least two cores.

It's recommended to smooth the input with filterFFT prior the detection.

The type of the return depends on the input parameters:

numeric (or a list of them) if width==1 & score==FALSE containing the position of the peaks.
data.frame (or list of them) if width==1 & score==TRUE containing a 'peak' column with the position of the peak plus a 'score' column with its score.
IRanges (or IRangesList) if width>1 & score==FALSE containing the ranges of the peaks.
GRanges if width>1 & score==TRUE containing the ranges of the peaks and the assigned score.

If width > 1, those ranges outside the range 1:length(data) will be skipped.

Oscar Flores oflores@mmb.pcb.ub.es

filterFFT(), peakScoring()

# Generate a random peaks profile
reads <- syntheticNucMap(nuc.len=40, lin.len=130)$syn.reads
cover <- coverage.rpm(reads)

# Filter them
cover_fft <- filterFFT(cover)

# Detect and plot peaks (up a bit the threshold for accounting synthetic
# data)
peaks <- peakDetection(cover_fft, threshold="40%", score=TRUE)
plotPeaks(peaks, cover_fft, threshold="40%", start=10000, end=15000)

# Now use ranges version, which accounts for fuzziness when scoring
peaks <- peakDetection(cover_fft, threshold="40%", score=TRUE, width=147)
plotPeaks(peaks, cover_fft, threshold="40%", start=10000, end=15000)