nCal (version 14.11-8)

crm.fit: Fit Concentration Response Model

Description

crm.fit can fit a constant or power variance function or log transform both sides.

Usage

crm.fit (formula, data, fit.4pl=FALSE, var.model=c("constant","power"),
    robust="mean", method=c("gls-pl","gnls","mle"), max.iter=50, reltol=1e-3, 
    gof.threshold=0.2, log.both.sides=FALSE, verbose=FALSE) 


## S3 method for class 'crm':
deviance(object, ...)
## S3 method for class 'crm':
lines(x, ...)
## S3 method for class 'crm':
coef(object, parameterization=c("cla","gh","ed50b","edb50"), ...)

Arguments

formula

data

fit.4pl

Boolean

var.model

string

robust

string

method

string

max.iter

number

reltol

numeric

gof.threshold

numeric

verbose

Boolean

log.both.sides

Boolean, log transform both sides

object, x

crm object

parameterization

string, output parameterization

...

additional argument

Value

• An object of crm and drm type.
• var.power
  estimated power parameter in the power variance function

Details

crm.fit implements an iterative method for estimating a model with power variance. method: gls-pl means GLS-PL (see reference) log.both.sides: transform both sides (see reference)

References

Fong, Y., Yu, X. (2014) Transformation Model Choice in Nonlinear Regression Analysis of Serial Dilution Assays, submitted

Examples

dat.std=dat.QIL3[dat.QIL3$assay_id=="LMX001",]

# run 3 iter to save time for examples
fit.1=crm.fit(fi~expected_conc, dat.std, var.model="power", verbose=TRUE, max.iter=2)
fit.2=crm.fit(log(fi)~expected_conc, dat.std, verbose=TRUE)
fit.3=crm.fit(log(fi)~expected_conc, dat.std, var.model="power", verbose=TRUE, max.iter=2)

sapply(list(fit.1, fit.2, fit.3), coef)
fit.1$var.power
fit.2$var.power
fit.3$var.power

plot(fit.1, log="xy", type="all", lwd=3, pch="*")
lines(fit.2, expy=TRUE, col=2, lwd=3)
lines(fit.3, expy=TRUE, col=4, lty=2, lwd=3)