fitCellGrowth: Fit growth curves
In SamGG/cellGrowth: Fitting cell population growth models

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

Fit a cell growth curve

1
2
3

  fitCellGrowth(x, z, model = "locfit",
    locfit.h = 3 * 60 * 60, locfit.deg = 2,
    relative.height.at.lag = 0.1)

`x`	`numeric` vector: time points
`z`	`numeric` vector: log(OD)
`model`	which model to fit.
`locfit.h`	`numeric`: `h` parameter (window size) in call to `locfit`. The default value is set to three hours assuming `x` given in seconds. You can detect a better bandwidth by calling `bandwidthCV`
`locfit.deg`	`numeric`: `deg` parameter (polynomials degree) in call to `locfit`
`relative.height.at.lag`	Parameter used by `guessCellGrowthParams`

For the non-parametric "locfit" model, local regression is done by a call to locfit. The returned maximum growth rate values the maximum value of the fitted derivative over the data points. For the parametric models "logistic", "gompertz", "rosso" and "baranyi", the function does a non-least square fit by calling nls. Initial parameters values are generated by guessCellGrowthParams. The returned maximum growth rate values the mu parameter of these models.

Fit as returned by locfit for the "locfit" model and as returned by nls for the "logistic", "gompertz", "rosso" and "baranyi" models. The returned value also has an attribute maxGrowthRate valueing the inferred maximum growth rate as well as an attribute pointOfMaxGrowthRate valuing the datapoint at which the growth rate is maximal. Also, it has an attribute max valuing the inferred maximum among the time points as well as pointOfMax valuing the datapoint of max fitted value. It gets the additional class cellCurveFit assigned.

Julien Gagneur and Moritz Matthey

nls, locfit, baranyi, gompertz, logistic, rosso, guessCellGrowthParams, fitCellGrowths

x = 1:1000
          z = gompertz(x, mu=0.01, l=200, z0=1, zmax=5) + rnorm(length(x),sd=0.1)
          f = fitCellGrowth(x, z, model = "gompertz")
          floc = fitCellGrowth(x, z, model = "locfit", locfit.h=500)
          	plot(x,z, main="simulated data\nGompertz model")
          	lines(x, predict(f,x), lwd=2, col="red")
          	lines(x, predict(floc,x), lwd=2, col="blue")
          	legend( "right", legend=c("gompertz fit", "locfit"), lwd=1, col=c("red","blue") )