zero.cross: Determine Extreme Values

In ornelles/flowExtra: Helper Functions for flowCore

Description

Determine minimum and maximum values from a smoothed curve through x-y points.

Usage

zero.cross(x, y, first = FALSE, spar = NULL)

Arguments

x, y	x- and y-values of the curve to search for extreme values
first	if TRUE, return only the first extreme value
spar	smoothing parameter for smooth.spline between 0 and 1

Details

Multiple x values are collapsed into a single value and the corresponding y value treated as the mean of multiple values. The resulting x and y values are smoothed by smooth.spline() with the smoothing parameter spar. If spar is NULL, the value will be determined by the algorithm and returned in model. This is the same code used in the ribofrag package although neither are exported.

Neither x nor y can have missing values.

The endpoints of the x, y points are included in the search for extreme values.

The function is called "zero.cross" because the point at which the first derivative crosses zero is used to determine the extreme. The fit and choice of spar can be evaluated by plot(x, y); lines(zero.cross(x, y)$model).

Value

A list identifying each extreme value found in the smoothed curve. The components of the list include x and y for the x- and y-values, sign which is +1 for a maximum and -1 for a minimum, and the smoothing model returned by smooth.spline() in model.

Examples

  set.seed(123)
  y_ex <- 100*cumsum(rnorm(250))
  x_ex <- seq_along(y_ex)
  ans <- zero.cross(x_ex, y_ex, spar = 0.5)
  as.data.frame(ans[1:3])
  plot(x_ex, y_ex)
  lines(ans$model)
  abline(v = ans$x, col = ifelse(ans$sign > 0, "black", NA))

  print(zero.cross(1:10, 1:10))

ornelles/flowExtra documentation built on March 1, 2020, 9:33 a.m.