ZIPLL: Zero-Inflated Piecewise Log-Logistic
In AnderWilson/ZIPLL: Fits the ZIPLL hierarchical dose-response model

Description Usage Arguments Value Author(s) References Examples

This implements the ZIPLL hierarchical dose-response estiamation.

1	ZIPLL(dat, nitter = 2000, burnin = 1000, internal.knots = 0, kappa = 1, mu = c(5, 1, 3), sigma = c(3, 0.1, 3))

`dat`	This is a matrix with columns (Chemical ID, Assay ID, Concentration, Response). Each column row corresponds to a unique dose-response pair for a single chemical-assay combination. The chemical and assay IDs should be integers.
`nitter`	Total number of MCMC itterations.
`burnin`	Number of itteration discarded for burnin. No data from these will be saved.
`internal.knots`	The location of internal knots. The default is one internal knot at 0.
`kappa`	This is the minimum difference between the Emin and Emax requried for a response to be considered active.
`mu`	The prior mean for the Emax, Emin, and AC50. Note that the AC50 is the transformed value.
`sigma`	The prior variance for the Emax, Emin, and AC50.

`dat`	This returns the data with the posterior predicted value and posterior standard deviation for each design point.
`parms`	This returns posterior mean and posterior variance for the AC50 and Emax and the probability of an active response. For chemicals-assay combinations that are not active the posterior mean and variance is drawing from the prior.

Ander Wilson

forthcoming

#simulate data
dat <- simdata(nchem=10, nassay=4, seed=1234)

#fit ZIPLL
fit <- ZIPLL(dat)

#plot a curve
dose<-fit$dat[,3]
response<-fit$dat[,4]
post.mn<-fit$dat[,5]
post.sd<-fit$dat[,6]

these<-dat[,1]==6 & dat[,2]==2
plot(dose[these],response[these],xlab="Log dose",ylab="Response", log="x")
lines(dose[these],post.mn[these])
lines(dose[these],post.mn[these]-2*post.sd[these],lty=2)
lines(dose[these],post.mn[these]+2*post.sd[these],lty=2)