mixed.mmkin and test coverage

7 files changed, 242 insertions, 5 deletions

diff --git a/tests/testthat/setup_script.R b/tests/testthat/setup_script.R
index 8d8ba3e9..9ec91425 100644
--- a/tests/testthat/setup_script.R
+++ b/tests/testthat/setup_script.R
@@ -100,3 +100,67 @@ fit_obs_1 <- mkinfit(m_synth_SFO_lin, SFO_lin_a, error_model = "obs", quiet = TR
 # We know threestep is OK, and threestep (and IRLS) is faster here
 fit_tc_1 <- mkinfit(m_synth_SFO_lin, SFO_lin_a, error_model = "tc", quiet = TRUE,
   error_model_algorithm = "threestep")
+
+# Mixed models data
+set.seed(123456)
+sampling_times = c(0, 1, 3, 7, 14, 28, 60, 90, 120)
+n <- n_biphasic <- 15
+log_sd <- 0.3
+err_1 = list(const = 1, prop = 0.05)
+tc <- function(value) sigma_twocomp(value, err_1$const, err_1$prop)
+const <- function(value) 2
+
+SFO <- mkinmod(parent = mkinsub("SFO"))
+k_parent = rlnorm(n, log(0.03), log_sd)
+ds_sfo <- lapply(1:n, function(i) {
+  ds_mean <- mkinpredict(SFO, c(k_parent = k_parent[i]),
+    c(parent = 100), sampling_times)
+  add_err(ds_mean, tc, n = 1)[[1]]
+})
+
+DFOP <- mkinmod(parent = mkinsub("DFOP"))
+dfop_pop <- list(parent_0 = 100, k1 = 0.06, k2 = 0.015, g = 0.4)
+dfop_parms <- as.matrix(data.frame(
+  k1 = rlnorm(n, log(dfop_pop$k1), log_sd),
+  k2 = rlnorm(n, log(dfop_pop$k2), log_sd),
+  g = plogis(rnorm(n, qlogis(dfop_pop$g), log_sd))))
+ds_dfop <- lapply(1:n, function(i) {
+  ds_mean <- mkinpredict(DFOP, dfop_parms[i, ],
+    c(parent = dfop_pop$parent_0), sampling_times)
+  add_err(ds_mean, const, n = 1)[[1]]
+})
+
+set.seed(123456)
+DFOP_SFO <- mkinmod(
+  parent = mkinsub("DFOP", "m1"),
+  m1 = mkinsub("SFO"),
+  quiet = TRUE)
+dfop_sfo_pop <- list(parent_0 = 100,
+  k_m1 = 0.002, f_parent_to_m1 = 0.5,
+  k1 = 0.05, k2 = 0.01, g = 0.5)
+syn_biphasic_parms <- as.matrix(data.frame(
+  k1 = rlnorm(n_biphasic, log(dfop_sfo_pop$k1), log_sd),
+  k2 = rlnorm(n_biphasic, log(dfop_sfo_pop$k2), log_sd),
+  g = plogis(rnorm(n_biphasic, qlogis(dfop_sfo_pop$g), log_sd)),
+  f_parent_to_m1 = plogis(rnorm(n_biphasic,
+      qlogis(dfop_sfo_pop$f_parent_to_m1), log_sd)),
+  k_m1 = rlnorm(n_biphasic, log(dfop_sfo_pop$k_m1), log_sd)))
+ds_biphasic_mean <- lapply(1:n_biphasic,
+  function(i) {
+    mkinpredict(DFOP_SFO, syn_biphasic_parms[i, ],
+      c(parent = 100, m1 = 0), sampling_times)
+  }
+)
+ds_biphasic <- lapply(ds_biphasic_mean, function(ds) {
+  add_err(ds,
+    sdfunc = function(value) sqrt(err_1$const^2 + value^2 * err_1$prop^2),
+    n = 1, secondary = "m1")[[1]]
+})
+
+# Mixed model fits
+mmkin_sfo_1 <- mmkin("SFO", ds_sfo, quiet = TRUE, error_model = "tc")
+sfo_saemix_1 <- saem(mmkin_sfo_1, quiet = TRUE, transformations = "saemix")
+mmkin_biphasic <- mmkin(list("DFOP-SFO" = DFOP_SFO), ds_biphasic, quiet = TRUE)
+nlme_biphasic <- nlme(mmkin_biphasic)
+saem_biphasic_m <- saem(mmkin_biphasic, transformations = "mkin", quiet = TRUE)
+saem_biphasic_s <- saem(mmkin_biphasic, transformations = "saemix", quiet = TRUE)
diff --git a/tests/testthat/test_AIC.R b/tests/testthat/test_AIC.R
index e9698f7c..57b9a673 100644
--- a/tests/testthat/test_AIC.R
+++ b/tests/testthat/test_AIC.R
@@ -6,6 +6,7 @@ test_that("The AIC is reproducible", {
                     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")
+  expect_error(BIC(fits["SFO", ]), "column object")
   expect_equivalent(BIC(fits[, "FOCUS_C"]),
                     data.frame(df = c(3, 4, 5, 5), AIC = c(59.9, 45.5, 30.0, 40.2)),
                     scale = 1, tolerance = 0.1)
diff --git a/tests/testthat/test_confidence.R b/tests/testthat/test_confidence.R
index 3fdd3f2c..54be675c 100644
--- a/tests/testthat/test_confidence.R
+++ b/tests/testthat/test_confidence.R
@@ -1,7 +1,20 @@
 context("Confidence intervals and p-values")
 
+test_that("Some special cases of confidence interval calculation work", {
+
+  tmp <- expect_warning(mkinfit("FOMC", FOCUS_2006_A, quiet = TRUE), "not converge")
+
+  expect_equivalent(
+    confint(tmp, transform = FALSE),
+    matrix(rep(NA, 8), nrow = 4))
+})
+
 test_that("The confint method 'quadratic' is consistent with the summary", {
   expect_equivalent(
+    confint(fit_nw_1, parm = "parent_0", method = "quadratic"),
+    summary(fit_nw_1)$bpar["parent_0", c("Lower", "Upper")])
+
+  expect_equivalent(
     confint(fit_nw_1, method = "quadratic"),
     summary(fit_nw_1)$bpar[, c("Lower", "Upper")])
 
@@ -74,8 +87,8 @@ test_that("Likelihood profile based confidence intervals work", {
    }
    f_mle <- stats4::mle(f_nll, start = as.list(parms(f)), nobs = nrow(FOCUS_2006_C))
 
-   ci_mkin_1_p_0.95 <- confint(f, method = "profile", level = 0.95,
-     cores = n_cores, quiet = TRUE)
+   ci_mkin_1_p_0.95 <- expect_message(confint(f, method = "profile", level = 0.95,
+     cores = n_cores, quiet = FALSE), "Profiling the likelihood")
 
    # Magically, we get very similar boundaries as stats4::mle
    # (we need to capture the output to avoid printing this while testing as
diff --git a/tests/testthat/test_mixed.R b/tests/testthat/test_mixed.R
new file mode 100644
index 00000000..2d69e13e
--- /dev/null
+++ b/tests/testthat/test_mixed.R
@@ -0,0 +1,35 @@
+context("Fitting of nonlinear mixed effects models")
+
+sampling_times = c(0, 1, 3, 7, 14, 28, 60, 90, 120)
+n_biphasic <- 8
+err_1 = list(const = 1, prop = 0.07)
+
+DFOP_SFO <- mkinmod(
+  parent = mkinsub("DFOP", "m1"),
+  m1 = mkinsub("SFO"),
+  quiet = TRUE)
+
+set.seed(123456)
+log_sd <- 0.3
+syn_biphasic_parms <- as.matrix(data.frame(
+  k1 = rlnorm(n_biphasic, log(0.05), log_sd),
+  k2 = rlnorm(n_biphasic, log(0.01), log_sd),
+  g = plogis(rnorm(n_biphasic, 0, log_sd)),
+  f_parent_to_m1 = plogis(rnorm(n_biphasic, 0, log_sd)),
+  k_m1 = rlnorm(n_biphasic, log(0.002), log_sd)))
+
+ds_biphasic_mean <- lapply(1:n_biphasic,
+  function(i) {
+    mkinpredict(DFOP_SFO, syn_biphasic_parms[i, ],
+      c(parent = 100, m1 = 0), sampling_times)
+  }
+)
+
+set.seed(123456L)
+ds_biphasic <- lapply(ds_biphasic_mean, function(ds) {
+  add_err(ds,
+    sdfunc = function(value) sqrt(err_1$const^2 + value^2 * err_1$prop^2),
+    n = 1, secondary = "m1")[[1]]
+})
+
+f_mmkin <- mmkin(list("DFOP-SFO" = DFOP_SFO), ds_biphasic, quiet = TRUE)
diff --git a/tests/testthat/test_nafta.R b/tests/testthat/test_nafta.R
index fcab4ffb..62c88983 100644
--- a/tests/testthat/test_nafta.R
+++ b/tests/testthat/test_nafta.R
@@ -20,7 +20,6 @@ test_that("Test data from Appendix B are correctly evaluated", {
 
   expect_known_output(print(res), "NAFTA_SOP_Appendix_B.txt")
 
-  skip_on_travis()
   plot_nafta <- function() plot(res)
   if(requireNamespace("vdiffr", quietly = TRUE)) {
     skip_if(getRversion() < "4.1.0")
@@ -49,7 +48,6 @@ test_that("Test data from Appendix D are correctly evaluated", {
 
   expect_known_output(print(res), "NAFTA_SOP_Appendix_D.txt")
 
-  skip_on_travis()
   plot_nafta <- function() plot(res)
   if(requireNamespace("vdiffr", quietly = TRUE)) {
     skip_if(getRversion() < "4.1.0")
diff --git a/tests/testthat/test_plot.R b/tests/testthat/test_plot.R
index 587ec02e..72f1020c 100644
--- a/tests/testthat/test_plot.R
+++ b/tests/testthat/test_plot.R
@@ -2,7 +2,6 @@ context("Plotting")
 
 test_that("Plotting mkinfit and mmkin objects is reproducible", {
   skip_on_cran()
-  skip_on_travis()
   plot_default_FOCUS_C_SFO <- function() plot(fits[["SFO", "FOCUS_C"]])
   plot_res_FOCUS_C_SFO <- function() plot(fits[["SFO", "FOCUS_C"]], show_residuals = TRUE)
   plot_res_FOCUS_C_SFO_2 <- function() plot_res(fits[["SFO", "FOCUS_C"]])
@@ -17,6 +16,11 @@ test_that("Plotting mkinfit and mmkin objects is reproducible", {
   plot_errmod_fit_D_obs_eigen <- function() plot_err(fit_D_obs_eigen, sep_obs = FALSE)
   plot_errmod_fit_C_tc <- function() plot_err(fit_C_tc)
 
+  plot_biphasic_mmkin <- function() plot(mixed(mmkin_biphasic))
+  plot_biphasic_nlme <- function() plot(nlme_biphasic)
+  plot_biphasic_saem_s <- function() plot(saem_biphasic_s)
+  plot_biphasic_saem_m <- function() plot(saem_biphasic_m)
+
   plot_res_sfo_sfo <- function() plot_res(f_sfo_sfo_desolve)
   plot_err_sfo_sfo <- function() plot_err(f_sfo_sfo_desolve)
   plot_errmod_fit_obs_1 <- function() plot_err(fit_obs_1, sep_obs = FALSE)
@@ -32,6 +36,10 @@ test_that("Plotting mkinfit and mmkin objects is reproducible", {
   vdiffr::expect_doppelganger("mmkin plot for FOCUS C", mmkin_FOCUS_C)
   vdiffr::expect_doppelganger("mmkin plot for SFO (FOCUS C and D)", mmkin_SFO)
   vdiffr::expect_doppelganger("plot_errmod with FOCUS C tc", plot_errmod_fit_C_tc)
+  vdiffr::expect_doppelganger("mixed model fit for mmkin object", plot_biphasic_mmkin)
+  vdiffr::expect_doppelganger("mixed model fit for nlme object", plot_biphasic_nlme)
+  vdiffr::expect_doppelganger("mixed model fit for saem object with saemix transformations", plot_biphasic_saem_s)
+  vdiffr::expect_doppelganger("mixed model fit for saem object with mkin transformations", plot_biphasic_saem_m)
   skip_on_travis() # Still not working on Travis, maybe because of deSolve producing
   # different results when not working with a compiled model or eigenvalues
   vdiffr::expect_doppelganger("plot_errmod with FOCUS D obs eigen", plot_errmod_fit_D_obs_eigen)
diff --git a/tests/testthat/test_saem.R b/tests/testthat/test_saem.R
new file mode 100644
index 00000000..0b6d4531
--- /dev/null
+++ b/tests/testthat/test_saem.R
@@ -0,0 +1,118 @@
+context("Nonlinear mixed effects models fitted with SAEM from saemix")
+
+set.seed(123456)
+sampling_times = c(0, 1, 3, 7, 14, 28, 60, 90, 120)
+n <- n_biphasic <- 15
+log_sd <- 0.3
+err_1 = list(const = 1, prop = 0.05)
+tc <- function(value) sigma_twocomp(value, err_1$const, err_1$prop)
+const <- function(value) 2
+
+SFO <- mkinmod(parent = mkinsub("SFO"))
+k_parent = rlnorm(n, log(0.03), log_sd)
+ds_sfo <- lapply(1:n, function(i) {
+  ds_mean <- mkinpredict(SFO, c(k_parent = k_parent[i]),
+    c(parent = 100), sampling_times)
+  add_err(ds_mean, tc, n = 1)[[1]]
+})
+
+DFOP <- mkinmod(parent = mkinsub("DFOP"))
+dfop_pop <- list(parent_0 = 100, k1 = 0.06, k2 = 0.015, g = 0.4)
+dfop_parms <- as.matrix(data.frame(
+  k1 = rlnorm(n, log(dfop_pop$k1), log_sd),
+  k2 = rlnorm(n, log(dfop_pop$k2), log_sd),
+  g = plogis(rnorm(n, qlogis(dfop_pop$g), log_sd))))
+ds_dfop <- lapply(1:n, function(i) {
+  ds_mean <- mkinpredict(DFOP, dfop_parms[i, ],
+    c(parent = dfop_pop$parent_0), sampling_times)
+  add_err(ds_mean, const, n = 1)[[1]]
+})
+
+set.seed(123456)
+DFOP_SFO <- mkinmod(
+  parent = mkinsub("DFOP", "m1"),
+  m1 = mkinsub("SFO"),
+  quiet = TRUE)
+syn_biphasic_parms <- as.matrix(data.frame(
+  k1 = rlnorm(n_biphasic, log(0.05), log_sd),
+  k2 = rlnorm(n_biphasic, log(0.01), log_sd),
+  g = plogis(rnorm(n_biphasic, 0, log_sd)),
+  f_parent_to_m1 = plogis(rnorm(n_biphasic, 0, log_sd)),
+  k_m1 = rlnorm(n_biphasic, log(0.002), log_sd)))
+ds_biphasic_mean <- lapply(1:n_biphasic,
+  function(i) {
+    mkinpredict(DFOP_SFO, syn_biphasic_parms[i, ],
+      c(parent = 100, m1 = 0), sampling_times)
+  }
+)
+ds_biphasic <- lapply(ds_biphasic_mean, function(ds) {
+  add_err(ds,
+    sdfunc = function(value) sqrt(err_1$const^2 + value^2 * err_1$prop^2),
+    n = 1, secondary = "m1")[[1]]
+})
+
+test_that("Parent only models can be fitted with saemix", {
+  # Some fits were done in the setup script
+  mmkin_sfo_2 <- mmkin("SFO", ds_sfo, fixed_initials = c(parent = 100), quiet = TRUE)
+
+  sfo_saemix_2 <- saem(mmkin_sfo_1, quiet = TRUE, transformations = "mkin")
+  sfo_saemix_3 <- expect_error(saem(mmkin_sfo_2, quiet = TRUE), "at least two parameters")
+  s_sfo_s1 <- summary(sfo_saemix_1)
+  s_sfo_s2 <- summary(sfo_saemix_2)
+
+  sfo_nlme_1 <- expect_warning(nlme(mmkin_sfo_1), "not converge")
+  s_sfo_n <- summary(sfo_nlme_1)
+
+  # Compare with input
+  expect_equal(round(s_sfo_s2$confint_ranef["SD.log_k_parent", "est."], 1), 0.3)
+  # k_parent is a bit different from input 0.03 here
+  expect_equal(round(s_sfo_s1$confint_back["k_parent", "est."], 3), 0.035)
+  expect_equal(round(s_sfo_s2$confint_back["k_parent", "est."], 3), 0.035)
+
+  # But the result is pretty unanimous between methods
+  expect_equal(round(s_sfo_s1$confint_back["k_parent", "est."], 3),
+    round(s_sfo_s2$confint_back["k_parent", "est."], 3))
+  expect_equal(round(s_sfo_s1$confint_back["k_parent", "est."], 3),
+    round(s_sfo_n$confint_back["k_parent", "est."], 3))
+
+  mmkin_dfop_1 <- mmkin("DFOP", ds_dfop, quiet = TRUE)
+
+  dfop_saemix_1 <- saem(mmkin_dfop_1, quiet = TRUE, transformations = "mkin")
+  dfop_saemix_2 <- saem(mmkin_dfop_1, quiet = TRUE, transformations = "saemix")
+  dfop_nlme_1 <- nlme(mmkin_dfop_1)
+  s_dfop_s1 <- summary(dfop_saemix_1)
+  s_dfop_s2 <- summary(dfop_saemix_2)
+  s_dfop_n <- summary(dfop_nlme_1)
+
+  dfop_pop <- as.numeric(dfop_pop)
+  expect_true(all(s_dfop_s1$confint_back[, "lower"] < dfop_pop))
+  expect_true(all(s_dfop_s1$confint_back[, "upper"] > dfop_pop))
+  expect_true(all(s_dfop_s2$confint_back[, "lower"] < dfop_pop))
+  expect_true(all(s_dfop_s2$confint_back[, "upper"] > dfop_pop))
+
+
+  # We get < 20% deviations with transformations made in mkin
+  rel_diff_1 <- (s_dfop_s1$confint_back[, "est."] - dfop_pop) / dfop_pop
+  expect_true(all(rel_diff_1 < 0.2))
+
+  # We get < 8% deviations with transformations made in saemix
+  rel_diff_2 <- (s_dfop_s2$confint_back[, "est."] - dfop_pop) / dfop_pop
+  expect_true(all(rel_diff_2 < 0.08))
+})
+
+test_that("Simple models with metabolite can be fitted with saemix", {
+
+  dfop_sfo_pop <- as.numeric(dfop_sfo_pop)
+  ci_dfop_sfo_s_s <- summary(saem_biphasic_s)$confint_back
+  expect_true(all(ci_dfop_sfo_s_s[, "lower"] < dfop_sfo_pop))
+  expect_true(all(ci_dfop_sfo_s_s[, "upper"] > dfop_sfo_pop))
+
+  # The following does not work, the k1 and k2 are not fitted well
+  ci_dfop_sfo_s_m <- summary(saem_biphasic_m)$confint_back
+  # expect_true(all(ci_dfop_sfo_s_m[, "lower"] < dfop_sfo_pop))
+  #expect_true(all(ci_dfop_sfo_s_m[, "upper"] > dfop_sfo_pop))
+
+  # Somehow this does not work at the moment. But it took forever (~ 10 min) anyways...
+  #saem_biphasic_2 <- saem(mmkin_biphasic, solution_type = "deSolve", quiet = TRUE)
+
+})