Improve tests, remove geometric_mean

9 files changed, 292 insertions, 72 deletions

diff --git a/tests/testthat/setup_script.R b/tests/testthat/setup_script.R
new file mode 100644
index 00000000..c1d46dcb
--- /dev/null
+++ b/tests/testthat/setup_script.R
@@ -0,0 +1,75 @@
+# Copyright (C) 2016-2019 Johannes Ranke
+# Contact: jranke@uni-bremen.de
+
+# This file is part of the R package mkin
+
+# mkin is free software: you can redistribute it and/or modify it under the
+# terms of the GNU General Public License as published by the Free Software
+# Foundation, either version 3 of the License, or (at your option) any later
+# version.
+
+# This program is distributed in the hope that it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+# FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+# details.
+
+# You should have received a copy of the GNU General Public License along with
+# this program. If not, see <http://www.gnu.org/licenses/>
+
+require(mkin)
+require(testthat)
+
+# Per default (on my box where I set NOT_CRAN) use all cores minus one
+if (identical(Sys.getenv("NOT_CRAN"), "true")) {
+  n_cores <- parallel::detectCores() - 1
+} else {
+  n_cores <- 1
+}
+
+# We are only allowed one core on travis, but they also set NOT_CRAN=true
+if (Sys.getenv("TRAVIS") != "") n_cores = 1
+
+# On Windows we would need to make a cluster first
+if (Sys.info()["sysname"] == "Windows") n_cores = 1
+
+# mmkin object of parent fits for tests
+models <- c("SFO", "FOMC", "DFOP", "HS")
+fits <- mmkin(models,
+  list(FOCUS_C = FOCUS_2006_C, FOCUS_D = FOCUS_2006_D),
+  quiet = TRUE, cores = n_cores)
+
+# One metabolite
+SFO_SFO <- mkinmod(parent = list(type = "SFO", to = "m1"),
+                   m1 = list(type = "SFO"), quiet = TRUE)
+SFO_SFO.ff <- mkinmod(parent = list(type = "SFO", to = "m1"),
+                      m1 = list(type = "SFO"),
+                      use_of_ff = "max", quiet = TRUE)
+
+f_sfo_sfo <- mkinfit(SFO_SFO,
+  subset(FOCUS_2006_D, value != 0),
+  quiet = TRUE)
+
+f_sfo_sfo.ff <- mkinfit(SFO_SFO.ff,
+  subset(FOCUS_2006_D, value != 0),
+  quiet = TRUE)
+
+# Two metabolites
+SFO_lin_a <- synthetic_data_for_UBA_2014[[1]]$data
+
+DFOP_par_c <- synthetic_data_for_UBA_2014[[12]]$data
+
+m_synth_SFO_lin <- mkinmod(
+  parent = mkinsub("SFO", "M1"),
+  M1 = mkinsub("SFO", "M2"),
+  M2 = mkinsub("SFO"),
+  use_of_ff = "max", quiet = TRUE)
+
+m_synth_DFOP_par <- mkinmod(parent = mkinsub("DFOP", c("M1", "M2")),
+  M1 = mkinsub("SFO"),
+  M2 = mkinsub("SFO"),
+  use_of_ff = "max", quiet = TRUE)
+
+f_SFO_lin_mkin_OLS <- mkinfit(m_synth_SFO_lin, SFO_lin_a, quiet = TRUE)
+f_SFO_lin_mkin_ML <- mkinfit(m_synth_SFO_lin, SFO_lin_a, quiet = TRUE,
+  error_model = "const", error_model_algorithm = "direct")
+
diff --git a/tests/testthat/slow/test_roundtrip_error_parameters.R b/tests/testthat/slow/test_roundtrip_error_parameters.R
index 97510563..7fa6c4e1 100644
--- a/tests/testthat/slow/test_roundtrip_error_parameters.R
+++ b/tests/testthat/slow/test_roundtrip_error_parameters.R
@@ -1,3 +1,36 @@
+# Copyright (C) 2018,2019 Johannes Ranke
+# Contact: jranke@uni-bremen.de
+
+# This file is part of the R package mkin
+
+# mkin is free software: you can redistribute it and/or modify it under the
+# terms of the GNU General Public License as published by the Free Software
+# Foundation, either version 3 of the License, or (at your option) any later
+# version.
+
+# This program is distributed in the hope that it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+# FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+# details.
+
+# You should have received a copy of the GNU General Public License along with
+# this program. If not, see <http://www.gnu.org/licenses/>
+
+context("Roundtripping error model parameters")
+
+# Per default (on my box where I set NOT_CRAN) use all cores minus one
+if (identical(Sys.getenv("NOT_CRAN"), "true")) {
+  n_cores <- parallel::detectCores() - 1
+} else {
+  n_cores <- 1
+}
+
+# We are only allowed one core on travis, but they also set NOT_CRAN=true
+if (Sys.getenv("TRAVIS") != "") n_cores = 1
+
+# On Windows we would need to make a cluster first
+if (Sys.info()["sysname"] == "Windows") n_cores = 1
+
 test_that("Reweighting method 'tc' produces reasonable variance estimates", {
 
   # Check if we can approximately obtain the parameters and the error model
@@ -19,18 +52,6 @@ test_that("Reweighting method 'tc' produces reasonable variance estimates", {
     sdfunc = function(x) sigma_twocomp(x, 0.5, 0.07),
     n = 1, reps = 100, digits = 5, LOD = -Inf, seed = 123456)
 
-  # Per default (on my box where I set NOT_CRAN) use all cores minus one
-  if (identical(Sys.getenv("NOT_CRAN"), "true")) {
-    n_cores <- parallel::detectCores() - 1
-  } else {
-    n_cores <- 1
-  }
-
-  # We are only allowed one core on travis, but they also set NOT_CRAN=true
-  if (Sys.getenv("TRAVIS") != "") n_cores = 1
-
-  # On Windows we would need to make a cluster first
-  if (Sys.info()["sysname"] == "Windows") n_cores = 1
 
   # Unweighted fits
   f_2_10 <- mmkin("DFOP", d_2_10, error_model = "const", quiet = TRUE,
diff --git a/tests/testthat/summary_DFOP_FOCUS_D.txt b/tests/testthat/summary_DFOP_FOCUS_D.txt
new file mode 100644
index 00000000..8abc15e7
--- /dev/null
+++ b/tests/testthat/summary_DFOP_FOCUS_D.txt
@@ -0,0 +1,125 @@
+mkin version used for fitting:    Dummy 0.0 for testing 
+R version used for fitting:       Dummy R version for testing 
+Date of fit:     Dummy date for testing 
+Date of summary: Dummy date for testing 
+
+Equations:
+d_parent/dt = - k_parent_sink * parent - k_parent_m1 * parent
+d_m1/dt = + k_parent_m1 * parent - k_m1_sink * m1
+
+Model predictions using solution type deSolve 
+
+Fitted using test 0 model solutions performed in test time 0 s
+
+Error model: Constant variance 
+
+Error model algorithm: OLS 
+
+Starting values for parameters to be optimised:
+                 value   type
+parent_0      100.7500  state
+k_parent_sink   0.1000 deparm
+k_parent_m1     0.1001 deparm
+k_m1_sink       0.1002 deparm
+
+Starting values for the transformed parameters actually optimised:
+                       value lower upper
+parent_0          100.750000  -Inf   Inf
+log_k_parent_sink  -2.302585  -Inf   Inf
+log_k_parent_m1    -2.301586  -Inf   Inf
+log_k_m1_sink      -2.300587  -Inf   Inf
+
+Fixed parameter values:
+     value  type
+m1_0     0 state
+
+Optimised, transformed parameters with symmetric confidence intervals:
+                  Estimate Std. Error  Lower   Upper
+parent_0            99.600    1.57000 96.400 102.800
+log_k_parent_sink   -3.038    0.07626 -3.193  -2.883
+log_k_parent_m1     -2.980    0.04033 -3.062  -2.898
+log_k_m1_sink       -5.248    0.13320 -5.518  -4.977
+sigma                3.126    0.35850  2.396   3.855
+
+Parameter correlation:
+                  parent_0 log_k_parent_sink log_k_parent_m1 log_k_m1_sink
+parent_0             1e+00             6e-01          -6e-02        -2e-01
+log_k_parent_sink    6e-01             1e+00          -9e-02        -6e-01
+log_k_parent_m1     -6e-02            -9e-02           1e+00         5e-01
+log_k_m1_sink       -2e-01            -6e-01           5e-01         1e+00
+sigma                1e-09            -9e-10           2e-08        -1e-10
+                   sigma
+parent_0           1e-09
+log_k_parent_sink -9e-10
+log_k_parent_m1    2e-08
+log_k_m1_sink     -1e-10
+sigma              1e+00
+
+Backtransformed parameters:
+Confidence intervals for internally transformed parameters are asymmetric.
+t-test (unrealistically) based on the assumption of normal distribution
+for estimators of untransformed parameters.
+               Estimate t value    Pr(>t)     Lower     Upper
+parent_0      99.600000  63.430 2.298e-36 96.400000 1.028e+02
+k_parent_sink  0.047920  13.110 6.126e-15  0.041030 5.596e-02
+k_parent_m1    0.050780  24.800 3.269e-23  0.046780 5.512e-02
+k_m1_sink      0.005261   7.510 6.165e-09  0.004012 6.898e-03
+sigma          3.126000   8.718 2.235e-10  2.396000 3.855e+00
+
+FOCUS Chi2 error levels in percent:
+         err.min n.optim df
+All data   6.398       4 15
+parent     6.827       3  6
+m1         4.490       1  9
+
+Resulting formation fractions:
+                ff
+parent_sink 0.4855
+parent_m1   0.5145
+m1_sink     1.0000
+
+Estimated disappearance times:
+          DT50   DT90
+parent   7.023  23.33
+m1     131.761 437.70
+
+Data:
+ time variable observed predicted   residual
+    0   parent    99.46  99.59848 -1.385e-01
+    0   parent   102.04  99.59848  2.442e+00
+    1   parent    93.50  90.23787  3.262e+00
+    1   parent    92.50  90.23787  2.262e+00
+    3   parent    63.23  74.07320 -1.084e+01
+    3   parent    68.99  74.07320 -5.083e+00
+    7   parent    52.32  49.91207  2.408e+00
+    7   parent    55.13  49.91207  5.218e+00
+   14   parent    27.27  25.01257  2.257e+00
+   14   parent    26.64  25.01257  1.627e+00
+   21   parent    11.50  12.53462 -1.035e+00
+   21   parent    11.64  12.53462 -8.946e-01
+   35   parent     2.85   3.14787 -2.979e-01
+   35   parent     2.91   3.14787 -2.379e-01
+   50   parent     0.69   0.71624 -2.624e-02
+   50   parent     0.63   0.71624 -8.624e-02
+   75   parent     0.05   0.06074 -1.074e-02
+   75   parent     0.06   0.06074 -7.382e-04
+    1       m1     4.84   4.80296  3.704e-02
+    1       m1     5.64   4.80296  8.370e-01
+    3       m1    12.91  13.02400 -1.140e-01
+    3       m1    12.96  13.02400 -6.400e-02
+    7       m1    22.97  25.04476 -2.075e+00
+    7       m1    24.47  25.04476 -5.748e-01
+   14       m1    41.69  36.69002  5.000e+00
+   14       m1    33.21  36.69002 -3.480e+00
+   21       m1    44.37  41.65310  2.717e+00
+   21       m1    46.44  41.65310  4.787e+00
+   35       m1    41.22  43.31312 -2.093e+00
+   35       m1    37.95  43.31312 -5.363e+00
+   50       m1    41.19  41.21831 -2.831e-02
+   50       m1    40.01  41.21831 -1.208e+00
+   75       m1    40.09  36.44703  3.643e+00
+   75       m1    33.85  36.44703 -2.597e+00
+  100       m1    31.04  31.98163 -9.416e-01
+  100       m1    33.13  31.98163  1.148e+00
+  120       m1    25.15  28.78984 -3.640e+00
+  120       m1    33.31  28.78984  4.520e+00
diff --git a/tests/testthat/test_CAKE_export.R b/tests/testthat/test_CAKE_export.R
index 23b424fc..a95a27ee 100644
--- a/tests/testthat/test_CAKE_export.R
+++ b/tests/testthat/test_CAKE_export.R
@@ -19,8 +19,9 @@
 context("Export dataset for reading into CAKE")
 
 test_that("Exporting is reproducible", {
-  CAKE_export(list("FOCUS C" = FOCUS_2006_C,
-                   "FOCUS D" = FOCUS_2006_D),
+  CAKE_export(
+    ds = list("FOCUS C" = FOCUS_2006_C,
+              "FOCUS D" = FOCUS_2006_D),
     map = c(parent = "Parent", m1 = "M1"),
     links = c(parent = "m1"),
     filename = "FOCUS_2006_D.csf", overwrite = TRUE, 
@@ -28,4 +29,7 @@ test_that("Exporting is reproducible", {
   csf <- readLines(con = "FOCUS_2006_D.csf")
   csf[8] <- "Date: Dummy date 0000-00-00"
   expect_known_value(csf, file = "FOCUS_2006_D.rds")
+  expect_error(CAKE_export(ds = list("FOCUS C" = FOCUS_2006_C),
+    filename = "FOCUS_2006_D.csf", overwrite = FALSE),
+    "already exists")
 })
diff --git a/tests/testthat/test_FOCUS_D_UBA_expertise.R b/tests/testthat/test_FOCUS_D_UBA_expertise.R
index a282f5e7..614305b1 100644
--- a/tests/testthat/test_FOCUS_D_UBA_expertise.R
+++ b/tests/testthat/test_FOCUS_D_UBA_expertise.R
@@ -19,12 +19,6 @@
 context("Results for FOCUS D established in expertise for UBA (Ranke 2014)")
 
 # Results are from p. 40
-SFO_SFO <- mkinmod(parent = list(type = "SFO", to = "m1"),
-                   m1 = list(type = "SFO"), quiet = TRUE)
-SFO_SFO.ff <- mkinmod(parent = list(type = "SFO", to = "m1"),
-                      m1 = list(type = "SFO"),
-                      use_of_ff = "max", quiet = TRUE)
-
 
 test_that("Fits without formation fractions are correct for FOCUS D", {
   fit.default <- expect_warning(mkinfit(SFO_SFO, FOCUS_2006_D, quiet = TRUE), "value of zero")
diff --git a/tests/testthat/test_SFORB.R b/tests/testthat/test_SFORB.R
new file mode 100644
index 00000000..49b3beed
--- /dev/null
+++ b/tests/testthat/test_SFORB.R
@@ -0,0 +1,32 @@
+# Copyright (C) 2019 Johannes Ranke
+# Contact: jranke@uni-bremen.de
+
+# This file is part of the R package mkin
+
+# mkin is free software: you can redistribute it and/or modify it under the
+# terms of the GNU General Public License as published by the Free Software
+# Foundation, either version 3 of the License, or (at your option) any later
+# version.
+
+# This program is distributed in the hope that it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+# FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+# details.
+
+# You should have received a copy of the GNU General Public License along with
+# this program. If not, see <http://www.gnu.org/licenses/>
+
+context("Fitting the SFORB model")
+
+logistic <- mkinmod(parent = mkinsub("logistic"))
+
+test_that("Fitting the SFORB model is equivalent to fitting DFOP", {
+  f_sforb <- mkinfit("SFORB", FOCUS_2006_C, quiet = TRUE)
+  f_dfop <- mkinfit("DFOP", FOCUS_2006_C, quiet = TRUE)
+  expect_equivalent(endpoints(f_sforb)$distimes, endpoints(f_dfop)$distimes,
+    tolerance = 1e-6)
+  s_sforb <- capture_output(print(summary(f_sforb)))
+  expect_match(s_sforb, "Estimated Eigenvalues of SFORB model\\(s\\):")
+  expect_match(s_sforb, "parent_b1 parent_b2")
+  expect_match(s_sforb, "0.45956 *0.01785")
+})
diff --git a/tests/testthat/test_confidence.R b/tests/testthat/test_confidence.R
index e5cc1954..0423302b 100644
--- a/tests/testthat/test_confidence.R
+++ b/tests/testthat/test_confidence.R
@@ -18,20 +18,8 @@
 
 context("Confidence intervals and p-values")
 
-m_synth_SFO_lin <- mkinmod(
-  parent = mkinsub("SFO", "M1"),
-  M1 = mkinsub("SFO", "M2"),
-  M2 = mkinsub("SFO"),
-  use_of_ff = "max", quiet = TRUE)
-
-SFO_lin_a <- synthetic_data_for_UBA_2014[[1]]$data
-
 test_that("Confidence intervals are stable", {
-  f_1_mkin_OLS <- mkinfit(m_synth_SFO_lin, SFO_lin_a, quiet = TRUE)
-  f_1_mkin_ML <- mkinfit(m_synth_SFO_lin, SFO_lin_a, quiet = TRUE,
-    error_model = "const", error_model_algorithm = "direct")
-
-  bpar_1 <- summary(f_1_mkin_ML)$bpar[, c("Estimate", "Lower", "Upper")]
+  bpar_1 <- summary(f_SFO_lin_mkin_ML)$bpar[, c("Estimate", "Lower", "Upper")]
   # The reference used here is mkin 0.9.48.1
   bpar_1_mkin_0.9 <-   read.table(text =
 "parent_0       102.0000 98.6000 106.0000
@@ -47,5 +35,6 @@ col.names = c("parameter", "estimate", "lower", "upper"))
   # Relative difference of lower bound of the confidence interval is < 0.02
   expect_equivalent(bpar_1[1:6, "Lower"], bpar_1_mkin_0.9$lower,
       scale = bpar_1_mkin_0.9$lower, tolerance = 0.02)
+  expect_equivalent(f_SFO_lin_mkin_OLS$bpar, f_SFO_lin_mkin_ML$bpar)
   })
 
diff --git a/tests/testthat/test_error_models.R b/tests/testthat/test_error_models.R
index f4015e00..ba40bae6 100644
--- a/tests/testthat/test_error_models.R
+++ b/tests/testthat/test_error_models.R
@@ -18,37 +18,6 @@
 
 context("Error model fitting")
 
-m_synth_SFO_lin <- mkinmod(parent = mkinsub("SFO", "M1"),
-                           M1 = mkinsub("SFO", "M2"),
-                           M2 = mkinsub("SFO"),
-                           use_of_ff = "max", quiet = TRUE)
-
-m_synth_DFOP_par <- mkinmod(parent = mkinsub("DFOP", c("M1", "M2")),
-                           M1 = mkinsub("SFO"),
-                           M2 = mkinsub("SFO"),
-                           use_of_ff = "max", quiet = TRUE)
-
-SFO_lin_a <- synthetic_data_for_UBA_2014[[1]]$data
-
-DFOP_par_c <- synthetic_data_for_UBA_2014[[12]]$data
-
-test_that("Error model 'const' works", {
-  skip_on_cran()
-  fit_const_1 <- mkinfit(m_synth_SFO_lin, SFO_lin_a, error_model = "const", quiet = TRUE)
-  bpar_1 <- fit_const_1$bparms.optim
-  # The reference used here is mkin 0.9.48.1
-  bpar_1_mkin_0.9 <-   read.table(text =
-"parent_0       102.0000
-k_parent         0.7390
-k_M1             0.2990
-k_M2             0.0202
-f_parent_to_M1   0.7690
-f_M1_to_M2       0.7230",
-col.names = c("parameter", "estimate"))
-
-  expect_equivalent(signif(bpar_1, 3), bpar_1_mkin_0.9$estimate)
-})
-
 test_that("Error model 'obs' works", {
   skip_on_cran()
   fit_obs_1 <- mkinfit(m_synth_SFO_lin, SFO_lin_a, error_model = "obs", quiet = TRUE)
diff --git a/tests/testthat/test_plots_summary_twa.R b/tests/testthat/test_plots_summary_twa.R
index 43036f1a..1dcf49ac 100644
--- a/tests/testthat/test_plots_summary_twa.R
+++ b/tests/testthat/test_plots_summary_twa.R
@@ -16,11 +16,6 @@
 # You should have received a copy of the GNU General Public License along with
 # this program. If not, see <http://www.gnu.org/licenses/>
 
-models <- c("SFO", "FOMC", "DFOP", "HS")
-fits <- mmkin(models,
-  list(FOCUS_C = FOCUS_2006_C, FOCUS_D = FOCUS_2006_D),
-  quiet = TRUE, cores = 1)
-
 context("Calculation of maximum time weighted average concentrations (TWAs)")
 
 test_that("Time weighted average concentrations are correct", {
@@ -49,7 +44,7 @@ test_that("Time weighted average concentrations are correct", {
 
 context("Summary")
 
-test_that("The summary is reproducible", {
+test_that("Summaries are reproducible", {
   fit <- fits[["DFOP", "FOCUS_C"]]
   test_summary <- summary(fit)
   test_summary$fit_version <- "Dummy 0.0 for testing"
@@ -63,6 +58,19 @@ test_that("The summary is reproducible", {
   # and between Travis and my own Linux system
   test_summary$Corr <- signif(test_summary$Corr, 1)
   expect_known_output(print(test_summary), "summary_DFOP_FOCUS_C.txt")
+
+  test_summary_2 <- summary(f_sfo_sfo,)
+  test_summary_2$fit_version <- "Dummy 0.0 for testing"
+  test_summary_2$fit_Rversion <- "Dummy R version for testing"
+  test_summary_2$date.fit <- "Dummy date for testing"
+  test_summary_2$date.summary <- "Dummy date for testing"
+  test_summary_2$calls <- "test 0"
+  test_summary_2$time <- c(elapsed = "test time 0")
+  # The correlation matrix is quite platform dependent
+  # It differs between i386 and amd64 on Windows
+  # and between Travis and my own Linux system
+  test_summary_2$Corr <- signif(test_summary_2$Corr, 1)
+  expect_known_output(print(test_summary_2), "summary_DFOP_FOCUS_D.txt")
 })
 
 context("Plotting")
@@ -71,11 +79,13 @@ test_that("Plotting mmkin objects is reproducible", {
   skip_on_cran()
   plot_sep_FOCUS_C_SFO <- function() plot_sep(fits[["SFO", "FOCUS_C"]])
   mkinparplot_FOCUS_C_SFO <- function() mkinparplot(fits[["SFO", "FOCUS_C"]])
+  mkinerrplot_FOCUS_C_SFO <- function() mkinerrplot(fits[["SFO", "FOCUS_C"]])
   mmkin_FOCUS_C <- function() plot(fits[, "FOCUS_C"])
   mmkin_SFO <- function() plot(fits["SFO",])
 
   vdiffr::expect_doppelganger("mkinfit plot for FOCUS C with sep = TRUE", plot_sep_FOCUS_C_SFO)
   vdiffr::expect_doppelganger("mkinparplot for FOCUS C SFO", mkinparplot_FOCUS_C_SFO)
+  vdiffr::expect_doppelganger("mkinerrplot for FOCUS C SFO", mkinerrplot_FOCUS_C_SFO)
   vdiffr::expect_doppelganger("mmkin plot for FOCUS C", mmkin_FOCUS_C)
   vdiffr::expect_doppelganger("mmkin plot for SFO (FOCUS C and D)", mmkin_SFO)
 })
@@ -84,7 +94,8 @@ context("AIC calculation")
 
 test_that("The AIC is reproducible", {
   expect_equivalent(AIC(fits[["SFO", "FOCUS_C"]]), 59.3, scale = 1, tolerance = 0.1)
-  expect_equivalent(AIC(fits[, "FOCUS_C"]), 
-                    data.frame(df = c(3, 4, 5, 5), AIC = c(59.3, 44.7, 29.0, 39.2)), 
+  expect_equivalent(AIC(fits[, "FOCUS_C"]),
+                    data.frame(df = c(3, 4, 5, 5), AIC = c(59.3, 44.7, 29.0, 39.2)),
                     scale = 1, tolerance = 0.1)
+  expect_error(AIC(fits["SFO", ]), "column object")
 })