updated NEWS and vignette

NEWS

@@ -2,7 +2,12 @@ aaresponse version 1.0.0 is described in
 R. Wehrens, J. Engel, J. Mes, A. de Jong and D. Esser, Amino
 Acids (2024) 56:25; https://doi.org/10.1007/s00726-024-03386-6.  
 
-Changes in version 1.0.2: (now develop branch)
+Changes in version 1.0.4:
+  - More flexible axis labels to allow the analysis of systems other
+    than amino acids
+  - Included two bug fixes by Guido Hooiveld
+  - Added "traditional" ways to estimate PoIs
+  - Accidentally skipped minor version numbers, too lazy to correct this
 
 Changes in version 1.0.1:
   - extractParameters now returns an object of class 'aar' rather than

R/compareInterventions.R

 fitMixedModels <-
   function(dt, refIntervention,
            model.formula, lm.alternative,
-           mainFun = lme4:::lmer, respondersOnly = TRUE)
+           mainFun = lme4::lmer, respondersOnly = TRUE)
 {
   imputedValues <- "AUC.orig" %in% colnames(dt)
 
@@ -110,7 +110,7 @@ compareInterventions <-
            logTransform = FALSE,
            model.formula = "~ Period + Intervention + (1 | Participant)",
            lm.alternative = "~ Period + Intervention",
-           mainFun = lme4:::lmer, singularFun = lm,
+           mainFun = lme4::lmer, singularFun = lm,
            respondersOnly = TRUE, ...)
 {
   model.terms <- trimws(strsplit(model.formula, "\\+")[[1]])

R/visualize.R

@@ -156,8 +156,7 @@ showDataFits <-
     for (pp in 1:nrow(params)) {
       if (!any(is.na(params[pp, parameters])))
           panel.lines(xx*15,
-                      aaresponse:::woodFun(params[pp, parameters],
-                                           xx - minTime/15),
+                      woodFun(params[pp, parameters], xx - minTime/15),
                     col = mycols[as.integer(params[pp, "Intervention"])])
     }
   }

man/compareInterventions.Rd

@@ -11,7 +11,7 @@ compareInterventions(
   logTransform = FALSE,
   model.formula = "~ Period + Intervention + (1 | Participant)",
   lm.alternative = "~ Period + Intervention",
-  mainFun = lme4:::lmer, singularFun = lm, respondersOnly = TRUE, ...)
+  mainFun = lme4::lmer, singularFun = lm, respondersOnly = TRUE, ...)
 }
 \arguments{
   \item{dt}{A data frame with at least the columns in the \code{target},

man/extractParameters.Rd

@@ -16,7 +16,7 @@ return the parameter of interest.
 \usage{
 getAUC(prs, maxT = 300)
 getHeight(prs)
-getTime2Max(prs, minutes = TRUE)
+getTime2Max(prs, minutes = TRUE, minTime = 0)
 extractParameters(prs.df, maxT = 300, minutes = TRUE)
 }
 \arguments{
@@ -28,12 +28,14 @@ extractParameters(prs.df, maxT = 300, minutes = TRUE)
     minutes.} 
   \item{minutes}{Whether the time to the peak maximum should be
     expressed in minutes. Default is TRUE.}
+  \item{minTime}{The earliest time point, which may be negative.}
 }
 \value{The simple extraction functions return scalars, function
   \code{extractParameters} returns an object of class 'aar'.}
 \author{Ron Wehrens}
 \seealso{
-  \code{\link{checkAAdata}}, \code{\link{extractParameters}}, \code{\link{showCIs}}
+  \code{\link{checkAAdata}}, \code{\link{fitWood}},
+  \code{\link{extractParameters}}, \code{\link{showCIs}}
 }
 \examples{
 \dontrun{

man/fitMixedModels.Rd

@@ -10,7 +10,7 @@
 \usage{
 fitMixedModels(dt, refIntervention,
                model.formula, lm.alternative,
-               mainFun = lme4:::lmer, respondersOnly = TRUE)
+               mainFun = lme4::lmer, respondersOnly = TRUE)
 }
 \arguments{
   \item{dt}{Data frame containing the relevant information.}

man/fitWood.Rd

@@ -5,7 +5,7 @@
 Fit a Wood curve through a time series.
 }
 \description{
-A Wood curve is given by y(t) = d + a t^{mc} e^{-ct} where d is the
+A Wood curve is given by \eqn{y(t) = d + a t^{mc} e^{-ct}} where d is the
 baseline, m is the time to the maximum, c is related to the rise and
 fall of the curve, and a to the height. The fit is done by numerical
 optimization using 500 random starts. In normal usage the \code{fitWood}
@@ -29,6 +29,11 @@ fitWoodAll(aadata, what = c("all", "aas", "essentials", "totals"))
   Participant, Period, Intervention and AA (using \code{aggregate}). Its
   result is an object of class "aar".
 }
+\details{Time values in Wood curves need to be non-negative. In
+  practice, sampling points may occur before t=0, the administration of
+  the treatment. Internally this is solved by simply setting the
+  earliest time point to zero for the curves, and shifting back in the
+  calculation of the parameters of interest, and in making figures.}
 \references{
 P.D.P. Wood, Algebraic model of the lactation curve in cattle. Nature
 216:164-165 (1967)

man/showCIs.Rd

@@ -9,9 +9,11 @@
   parameters of interest, AUC, peak height, and time to peak maximum. 
 }
 \usage{
-showCIs(resultsTable, ...)
+showCIs(resultsTable,
+        ylab = expression(paste("AA level (", mu, "M)", sep = "")), ...)
 showCombinedCIs(lAUC = NULL, AUC = NULL, Height = NULL, Time2Max = NULL,
-                relevantAAs, what, between, subset, ...)
+                relevantAAs, what = c("all", "aas", "essentials", "totals"),
+                between, subset, ylab = "Amino acid", ...)
 }
 \arguments{
   \item{resultsTable, lAUC, AUC, Height, Time2Max}{An object generated
@@ -29,7 +31,7 @@ showCombinedCIs(lAUC = NULL, AUC = NULL, Height = NULL, Time2Max = NULL,
     results on a ratio scale from the other PoIs that are always on a
     difference scale.}
   \item{subset}{Optional definition of a subset of amino acids.}
-  \item{\dots}{Additional arguments to the underlying plot function,
+  \item{ylab,\dots}{Additional arguments to the underlying plot function,
     most often used to define a subset of the results to be displayed
     (see examples).}
 }

vignettes/aaresponse.Rnw

@@ -477,6 +477,16 @@ starting points and quite generous optimization boundaries to avoid
 missing the true optimum. The result is a named vector, also containing
 the RMS value of the fit.
 
+As of version 1.0.4 of the package, it is possible to have time points
+that are sampled before $t = 0$ -- this is a very useful strategy
+since it adds to the reliability of the estimates of the
+baselines. The Wood curves, however, require non-negative time
+values. Therefore, in cases with negative time points, the time axis is
+shifted in the fitting of the Wood curves so that the first time point
+equals $t = 0$. Again, this means that the Wood curve parameters need
+to be interpreted on a different scale than the data. This is relevant
+in plotting, for example, and in extracting PoIs (see below).
+
 Function \code{fitWood} is useful to assess individual cases, but will
 rarely be used directly -- in
 principle one will want to fit all curves in a data set. This can