simlogdr: Simulate log dose-response data

In TJMurphy/nlfitr: Fitting models to nonlinear data

Description

A sandbox to create and plot replicate response data with random normal error on a log scale. Enter a log scaled predictor variable and values for additional model parameter arguments. The nonlinear model formula is derived from the general hyperbolic stimulus-response function: y/ymax=x^h/(x^h+k^h), where ymax = yhi - ylo. Errors in geom_smooth fitting will occasionally happen. Just re-simulate or modify parameters. The data generating equation is: 'y = ylo + (yhi - ylo)/(1 + 10^((logk - x)*h)) + rnorm(length(x), 0, sd)' The regression formula is: 'y ~ ylo + (yhi - ylo)/(1 + 10^((logk - x)*h))“

Usage

simlogdr(x, logk, ylo, yhi, h, sd, reps)

Arguments

x	a vector of log scale values; usually log10 or log2 dose or concentration units.
logk	the value of x in log units that yields y/ymax = 0.5; usually logEC50 or logED50.
ylo	the lowest expected y value, in response units.
yhi	the highest expected y value, in response units.
h	the Hill slope, a unitless slope factor; -1 > h > 1 is steeper, -1 < h < 1 is shallower. Provide negative value for downward sloping response.
sd	the standard deviation of residual error, in response units.
reps	and integer value for number of replicates.

Value

ggplot, data

Examples

# x is equivalent to log10(c(1e-10, 3e-10, 1e-9, 3e-9, 1e-8, 3e-8, 1e-7, 3e-7, 1e-6)).
# note these yield approximately half unit spacing on log10 scale

simUp <- simlogdr(x = c(-10, -9.523, -9, -8.523, -8, -7.523, -7, -6.523, -6),
               logk = -8, ylo = 300, yhi = 3000,
               h = 1.0, sd = 100, reps = 5); simUp

# grab simulated data for other uses

simUp$data

# use negative values of h to simulate downward sloping responses

simDown <- simlogdr(x = c(-10, -9.523, -9, -8.523, -8, -7.523, -7, -6.523, -6),
               logk = -8, ylo = 300, yhi = 3000,
               h = -2.5, sd = 100, reps = 5); simDown

TJMurphy/nlfitr documentation built on March 18, 2021, 12:33 p.m.