R/find_peaks.R

In GCalcium: A Data Manipulation and Analysis Package for Calcium Indicator Data

#' @title Get peaks/valleys in waveform data
#'
#' @description `find_peaks` finds peaks or valleys in waveform by using inflection points, with filter of 'n' increasing/decreasing points on both sides of each inflection point.
#'
#' @param x vector of numbers
#' @param n.points  the number of decreasing (peaks) or increasing (valleys) data points on left and right of inflection point required to be considered a "peak". A positive number as an input finds peaks, and a negative number finds valleys.
#' @return a numeric vector of indices
#' @examples
#' ### Format data frame
#' df.new <- format_data(GCaMP)
#'
#' ### How many peaks are there in trial 1 with 10 decreasing data points on each side?
#' peak.indices <- find_peaks(df.new$Trial1, n.points = 10)
#'
#' ### When do they occur?
#' data.frame(times = df.new$Time[peak.indices], vals = df.new$Trial1[peak.indices])
#' @export

find_peaks <- function (x, n.points = 3){

  # define curves and peaks
  curves <- diff(sign(diff(x, na.pad = FALSE)))
  peakz <- sapply(which(curves < 0), FUN = function(i){

    b <- i - n.points + 1
    b <- ifelse(b > 0, b, 1)
    a <- i + n.points + 1
    a <- ifelse(a < length(x), a, length(x))
    if(all(x[c(b : i, (i + 2) : a)] <= x[i + 1])) return(i + 1) else return(numeric(0))

  })

  peakz <- unlist(peakz)
  return(peakz)

}