Merge pull request #17 from weberse2/issue-rel-173

1.7-3 release

.Rbuildignore

@@ -33,3 +33,4 @@ src/stan_files/.*\.hpp$
 ^src/package-binary$
 ^Makefile$
 ^LICENSE$
+^vignettes/articles$

DESCRIPTION

@@ -8,8 +8,8 @@ Description: Tool-set to support Bayesian evidence synthesis.  This
     for details on applying this package while Neuenschwander et al. (2010)
     <doi:10.1177/1740774509356002> and Schmidli et al. (2014)
     <doi:10.1111/biom.12242> explain details on the methodology.
-Version: 1.7-2
-Date: 2023-08-21
+Version: 1.7-3
+Date: 2024-01-02
 Authors@R: c(person("Novartis", "Pharma AG", role = "cph")
             ,person("Sebastian", "Weber", email="sebastian.weber@novartis.com", role=c("aut", "cre"))
             ,person("Beat", "Neuenschwander", email="beat.neuenschwander@novartis.com", role="ctb")
@@ -25,7 +25,7 @@ Imports:
     methods,
     Rcpp (>= 0.12.0),
     RcppParallel (>= 5.0.1),
-    rstan (>= 2.19.3),
+    rstan (>= 2.26.0),
     rstantools (>= 2.3.1),
     assertthat,
     mvtnorm,
@@ -44,8 +44,8 @@ LinkingTo:
     Rcpp (>= 0.12.0),
     RcppEigen (>= 0.3.3.3.0),
     RcppParallel (>= 5.0.1),
-    rstan (>= 2.19.3),
-    StanHeaders (>= 2.19.0)
+    rstan (>= 2.26.0),
+    StanHeaders (>= 2.26.0)
 License: GPL (>=3)
 LazyData: true
 Biarch: true
@@ -69,6 +69,6 @@ Suggests:
     glue,
     ragg
 VignetteBuilder: knitr
-SystemRequirements: GNU make, pandoc (>= 1.12.3), pandoc-citeproc, C++17
+SystemRequirements: GNU make, pandoc (>= 1.12.3), pngquant, C++17
 Encoding: UTF-8
 RoxygenNote: 7.2.3

NEWS.md

@@ -1,3 +1,18 @@
+# RBesT 1.7-3 - January 2nd, 2024
+
+## Enhancements
+
+* Updated Stan model file syntax to use new array syntax as required
+  by Stan >=2.33. This upgrades the minimal Stan version to 2.26.
+* Moved most vignettes to be articles to decrease size of R
+  packages. Articles are available on [the package
+  homepage](https://opensource.nibr.com/RBesT/articles/).
+
+## Bugfixes
+
+* Added `pngquant` to system requirements of package as requested by
+  CRAN.
+
 # RBesT 1.7-2 - August 21st, 2023
 
 ## Enhancements

R/mixplot.R

@@ -92,7 +92,7 @@ plot.mix <- function(x, prob=0.99, fun=dmix, log=FALSE, comp=TRUE, size=1.25, ..
 
     num_comp <- ncol(x)
     pl <- ggplot(data.frame(x=interval), aes(x=x)) +
-        stat_function(geom=plot_geom, fun = plot_fun, args=list(mix=x, log=log), n=n_fun, size=size) +
+        stat_function(geom=plot_geom, fun = plot_fun, args=list(mix=x, log=log), n=n_fun, linewidth=size) +
         bayesplot::bayesplot_theme_get()
 
     if(funStr=="dmix") {

inst/stan/gMAP.stan

@@ -14,22 +14,22 @@ data {
   vector[H] y_se;
 
   // binomial data, link=logit=2
-  int<lower=0>    r[H];
-  int<lower=1>  r_n[H];
+  array[H] int<lower=0>    r;
+  array[H] int<lower=1>  r_n;
 
   // count data, link=log=3
-  int<lower=0> count[H];
+  array[H] int<lower=0> count;
   vector[H]    log_offset;
 
   // exchangeability cluster mapping
   int<lower=1> n_groups;
-  int<lower=1,upper=n_groups> group_index[H];
+  array[H] int<lower=1,upper=n_groups> group_index;
 
   // tau prediction stratum
   int<lower=1,upper=n_groups> n_tau_strata;
   int<lower=1,upper=n_tau_strata> tau_strata_pred;
   // data item to tau stratum mapping
-  int<lower=1,upper=n_tau_strata> tau_strata_index[H];
+  array[H] int<lower=1,upper=n_tau_strata> tau_strata_index;
 
   // number of predictors
   int<lower=1> mX;
@@ -52,20 +52,20 @@ data {
   int<lower=0,upper=1> ncp;
 
   // guesses on the parameter location and scales
-  vector[mX] beta_raw_guess[2];
-  real       tau_raw_guess[2];
+  array[2] vector[mX] beta_raw_guess;
+  array[2] real       tau_raw_guess;
 
   // sample from prior predictive (do not add data to likelihood)
   int<lower=0,upper=1> prior_PD;
 }
 transformed data {
-  vector[mX] beta_prior_stan[2];
-  vector[n_tau_strata] tau_prior_stan[2];
+  array[2] vector[mX] beta_prior_stan;
+  array[2] vector[n_tau_strata] tau_prior_stan;
   //matrix[n_groups, n_tau_strata] S;
   //matrix[H, n_groups] Z;
   matrix[H, mX] X_param;
   // group index to tau stratum mapping
-  int<lower=1,upper=n_tau_strata> tau_strata_gindex[n_groups] = rep_array(tau_strata_pred, n_groups);
+  array[n_groups] int<lower=1,upper=n_tau_strata> tau_strata_gindex = rep_array(tau_strata_pred, n_groups);
 
   for (i in 1:mX) {
     beta_prior_stan[1,i] = beta_prior[i,1];

tests/testthat/test-gMAP.R

@@ -104,6 +104,7 @@ test_that("SBC data was up to date at package creation.", {
 set.seed(92575)
 rate <- round(-log(0.05)/2, 1)
 test_that("gMAP matches RStanArm binomial family", {
+    skip("RStanArm has issues loading since 2024-01-02 on CI/CD systems.")
               skip_on_cran()
               best_run <- gMAP(cbind(r, n-r) ~ 1 | study,
                                data=AS,

tools/make-ds.R

@@ -43,7 +43,7 @@ make_internal_ds <- function() {
     calibration_meta["MD5"]  <- vals["MD5"]
 
     pkg_create_date  <- Sys.time()
-    pkg_sha <- "0e02ce3"
+    pkg_sha <- "511a0f1"
 
     if (gsub("\\$", "", pkg_sha) == "Format:%h") {
         pkg_sha <- system("git rev-parse --short HEAD", intern=TRUE)

vignettes/.gitignore

@@ -0,0 +1,2 @@
+*.html
+*.R

vignettes/PoS_codata.Rmd → vignettes/articles/PoS_codata.Rmd

@@ -24,9 +24,9 @@ vignette: >
   %\VignetteEngine{knitr::rmarkdown}
 ---
 
-```{r, child="settings-knitr.txt"}
+```{r, child="../settings-knitr.txt"}
 ```
-```{r, child="settings-sampling.txt"}
+```{r, child="../settings-sampling.txt"}
 ```
 ```{r, include=FALSE}
 library(bayesplot)

vignettes/PoS_interim.Rmd → vignettes/articles/PoS_interim.Rmd

@@ -9,9 +9,9 @@ vignette: >
   %\VignetteEngine{knitr::rmarkdown}
 ---
 
-```{r, child="settings-knitr.txt"}
+```{r, child="../settings-knitr.txt"}
 ```
-```{r, child="settings-sampling.txt"}
+```{r, child="../settings-sampling.txt"}
 ```
 ```{r init, include=FALSE}
 library(scales)

vignettes/customizing_plots.Rmd → vignettes/articles/customizing_plots.Rmd

@@ -9,9 +9,9 @@ vignette: >
   %\VignetteEngine{knitr::rmarkdown}
 ---
 
-```{r, child="settings-knitr.txt"}
+```{r, child="../settings-knitr.txt"}
 ```
-```{r, child="settings-sampling.txt"}
+```{r, child="../settings-sampling.txt"}
 ```
 
 # Introduction

vignettes/introduction_normal.Rmd → vignettes/articles/introduction_normal.Rmd

@@ -9,9 +9,9 @@ vignette: >
   %\VignetteEngine{knitr::rmarkdown}
 ---
 
-```{r, child="settings-knitr.txt"}
+```{r, child="../settings-knitr.txt"}
 ```
-```{r, child="settings-sampling.txt"}
+```{r, child="../settings-sampling.txt"}
 ```
 
 # Introduction

vignettes/variances_MAP.Rmd → vignettes/articles/variances_MAP.Rmd

@@ -10,9 +10,9 @@ vignette: >
 ---
 
 
-```{r, child="settings-knitr.txt"}
+```{r, child="../settings-knitr.txt"}
 ```
-```{r, child="settings-sampling.txt"}
+```{r, child="../settings-sampling.txt"}
 ```
 ```{r, include=FALSE}
 library(dplyr)

vignettes/introduction.Rmd

@@ -30,17 +30,30 @@ vignette: >
 
 # Introduction
 
-The R Bayesian evidence synthesis Tools (RBesT) have been created to
-facilitate the use of historical information in clinical trials. Once
-relevant historical information has been identified, RBesT supports
-the derivation of informative priors via the Meta-Analytic-Predictive
-(MAP) approach [1] and the evaluation of the trial's operating
-characteristics. The MAP approach performs a standard meta-analysis
-followed by a prediction for the control group parameter of a future
-study while accounting for the uncertainty in the population mean (the
-standard result from a meta-analysis) and the between-trial
-heterogeneity. Therefore, RBesT can also be used as a meta-analysis
-tool if one simply neglects the prediction part.
+The R Bayesian evidence synthesis Tools (RBesT) facilitate the use of
+historical information in clinical trials. Once relevant historical
+information has been identified, RBesT supports the derivation of
+informative priors via the Meta-Analytic-Predictive (MAP) approach [1]
+and the evaluation of the trial's operating characteristics. The MAP
+approach performs a standard meta-analysis followed by a prediction
+for the control group parameter of a future study while accounting for
+the uncertainty in the population mean (the standard result from a
+meta-analysis) and the between-trial heterogeneity. Therefore, RBesT
+can also be used as a meta-analysis tool if one simply neglects the
+prediction part.
+
+This document demonstrates RBesT as it can be used to derive from
+historical control data a prior for a binary endpoint. The [RBesT
+package homepage](https://opensource.nibr.com/RBesT/) contains further
+articles on introductory material:
+
+- Probability of success with co-data
+- Probability of success at an interim with a normal endpoint
+- Customizing RBesT plots
+- RBesT for a normal endpoint
+- Meta-Analytic-Predictive priors for variances
+
+# Binary responder analysis example
 
 Let's consider a Novartis Phase II study in ankylosing spondylitis
 comparing the Novartis test treatment secukinumab with placebo