bmdboot: Computation of confidence interval on benchmark doses by...

In DRomics: Dose Response for Omics

Computation of confidence interval on benchmark doses by bootstrap

Description

Computes 95 percent confidence intervals on x-fold and z-SD benchmark doses by bootstrap.

Usage

bmdboot(r, items = r$res$id, niter = 1000, 
                    conf.level = 0.95, 
                    tol = 0.5, progressbar = TRUE, 
                    parallel = c("no", "snow", "multicore"), ncpus)

## S3 method for class 'bmdboot'
print(x, ...)

## S3 method for class 'bmdboot'
plot(x, BMDtype = c("zSD", "xfold"), remove.infinite = TRUE,
                   by = c("none", "trend", "model", "typology"), 
                   CI.col = "blue", BMD_log_transfo = TRUE,  ...) 

Arguments

r	An object of class "bmdcalc" returned by the function bmdcalc.
items	A character vector specifying the identifiers of the items for which you want the computation of confidence intervals. If omitted the computation is done for all the items.
niter	The number of samples drawn by bootstrap.
conf.level	Confidence level of the intervals.
tol	The tolerance in term of proportion of bootstrap samples on which the fit of the model is successful (if this proportion is below the tolerance, NA values are given for the limits of the confidence interval.
progressbar	If TRUE a progress bar is used to follow the bootstrap process.
parallel	The type of parallel operation to be used, "snow" or "multicore" (the second one not being available on Windows), or "no" if no parallel operation.
ncpus	Number of processes to be used in parallel operation : typically one would fix it to the number of available CPUs.
x	An object of class "bmdboot".
BMDtype	The type of BMD to plot, "zSD" (default choice) or "xfold".
remove.infinite	If TRUE the confidence intervals with non finite upper bound are not plotted.
by	If not at "none" the plot is split by the indicated factor ("trend", "model" or "typology").
CI.col	The color to draw the confidence intervals.
BMD_log_transfo	If TRUE, default option, a log transformation of the BMD is used in the plot.
...	Further arguments passed to graphical or print functions.

Details

Non-parametric bootstrapping is used, where mean centered residuals are bootstrapped. For each item, bootstrapped parameter estimates are obtained by fitting the model on each of the resampled data sets. If the fitting procedure fails to converge in more than tol*100% of the cases, NA values are given for the confidence interval. Otherwise, bootstraped BMD are computed from bootstrapped parameter estimates using the same method as in bmdcalc. Confidence intervals on BMD are then computed using percentiles of the bootstrapped BMDs. For example 95 percent confidence intervals are computed using 2.5 and 97.5 percentiles of the bootstrapped BMDs. In cases where the bootstrapped BMD cannot be estimated as not reached at the highest tested dose or not reachable due to model asymptotes, it was given an infinite value Inf, so as to enable the computation of the lower limit of the BMD confidence interval if a sufficient number of bootstrapped BMD values were estimated to finite values.

Value

bmdboot returns an object of class "bmdboot", a list with 3 components:

res	a data frame reporting the results of the fit, BMD computation and bootstrap on each specified item sorted in the ascending order of the adjusted p-values. The different columns correspond to the identifier of each item (id), the row number of this item in the initial data set (irow), the adjusted p-value of the selection step (adjpvalue), the name of the best fit model (model), the number of fitted parameters (nbpar), the values of the parameters b, c, d, e and f, (NA for non used parameters), the residual standard deviation (SDres), the typology of the curve (typology, (16 class typology described in the help of the drcfit function)), the rough trend of the curve (trend) defined with four classes (U, bell, increasing or decreasing shape), the theoretical y value at the control (y0), the theoretical y value at the maximal dose yatdosemax), the theoretical y range for x within the range of tested doses (yrange), the maximal absolute y change (up or down) from the control(maxychange) and for biphasic curves the x value at which their extremum is reached (xextrem) and the corresponding y value (yextrem), the BMD-zSD value (BMD.zSD) with the corresponding BMR-zSD value (reached or not, BMR.zSD) and the BMD-xfold value (BMD.xfold) with the corresponding BMR-xfold value (reached or not, BMR.xfold), BMD.zSD.lower and BMD.zSD.upper the lower and upper bounds of the confidence intervals of the BMD-zSD value, BMD.xfold.lower and BMD.xfold.upper the lower and upper bounds of the confidence intervals of the BMD-xfold value and nboot.successful the number of successful fits on bootstrapped samples for each item.
z	Value of z given in input to define the BMD-zSD.
x	Value of x given in input as a percentage to define the BMD-xfold.
tol	The tolerance given in input in term of tolerated proportion of failures of fit on bootstrapped samples.
niter	The number of samples drawn by bootstrap (given in input).

Author(s)

Marie-Laure Delignette-Muller

References

Huet S, Bouvier A, Poursat M-A, Jolivet E (2003) Statistical tools for nonlinear regression: a practical guide with S-PLUS and R examples. Springer, Berlin, Heidelberg, New York.

See Also

See bmdcalc for details about the computation of benchmark doses.

Examples

# (1) a toy example (a very small subsample of a microarray data set) 
#
datafilename <- system.file("extdata", "transcripto_very_small_sample.txt",
   package = "DRomics")

# to test the package on a small but not very small data set
# use the following commented line
# datafilename <- system.file("extdata", "transcripto_sample.txt", package = "DRomics")

o <- microarraydata(datafilename, check = TRUE, norm.method = "cyclicloess")
s_quad <- itemselect(o, select.method = "quadratic", FDR = 0.001)
f <- drcfit(s_quad, progressbar = TRUE)
r <- bmdcalc(f)
set.seed(1234) # to get reproducible results with a so small number of iterations
(b <- bmdboot(r, niter = 5)) # with a non reasonable value for niter 
# !!!! TO GET CORRECT RESULTS
# !!!! niter SHOULD BE FIXED FAR LARGER , e.g. to 1000 
# !!!! but the run will be longer 
b$res
plot(b) # plot of BMD.zSD after removing of BMDs with infinite upper bounds


# same plot in raw scale (without log transformation of BMD values)
plot(b, BMD_log_transfo = FALSE)

# plot of BMD.zSD without removing of BMDs 
# with infinite upper bounds
plot(b, remove.infinite = FALSE) 



# bootstrap on only a subsample of items
# with a greater number of iterations

chosenitems <- r$res$id[1:5] 
(b.95 <- bmdboot(r, items = chosenitems,
                     niter = 1000, progressbar = TRUE))
b.95$res

# Plot of fits with BMD values  and confidence intervals
# with the default BMD.zSD
plot(f, items = chosenitems, BMDoutput = b.95, BMDtype = "zSD")
# with the default BMD.xfold 
plot(f, items = chosenitems, BMDoutput = b.95, BMDtype = "xfold")

# same bootstrap but changing the default confidence level (0.95) to 0.90
(b.90 <- bmdboot(r, items = chosenitems,
                       niter = 1000, conf.level = 0.9, progressbar = TRUE))
b.90$res






# (2) an example on a microarray data set (a subsample of a greater data set) 
#

datafilename <- system.file("extdata", "transcripto_sample.txt", package="DRomics")

(o <- microarraydata(datafilename, check = TRUE, norm.method = "cyclicloess"))
(s_quad <- itemselect(o, select.method = "quadratic", FDR = 0.001))
(f <- drcfit(s_quad, progressbar = TRUE))
(r <- bmdcalc(f))
(b <- bmdboot(r, niter = 100)) # niter to put at 1000 for a better precision

# different plots of BMD-zSD
plot(b)
plot(b, by = "trend") 
plot(b, by = "model") 
plot(b, by = "typology") 

# a plot of BMD-xfold (by default BMD-zSD is plotted)
plot(b, BMDtype = "xfold") 


# (3) Comparison of parallel and non parallel implementations 
#

# to be tested with a greater number of iterations
if(!requireNamespace("parallel", quietly = TRUE)) {
   if(parallel::detectCores() > 1) {
      system.time(b1 <- bmdboot(r, niter = 100, progressbar = TRUE))
      system.time(b2 <- bmdboot(r, niter = 100, progressbar = FALSE, parallel = "snow", ncpus = 2))
}}

DRomics documentation built on Oct. 16, 2024, 5:09 p.m.