From dddb52ccb0f67d1dd97c1c09dd40812abebc49b1 Mon Sep 17 00:00:00 2001
From: Jack Leary <j.leary@ufl.edu>
Date: Sat, 19 Oct 2024 09:51:46 -0400
Subject: [PATCH] fixed a bug in GLMM RE bootstrapper

---
 R/bootstrapRandomEffects.R   | 19 +++++++++++++++++++
 tests/testthat/test_scLANE.R |  8 ++++----
 2 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/R/bootstrapRandomEffects.R b/R/bootstrapRandomEffects.R
index 2fdac60..b826e4d 100644
--- a/R/bootstrapRandomEffects.R
+++ b/R/bootstrapRandomEffects.R
@@ -113,10 +113,29 @@ bootstrapRandomEffects <- function(glmm.mod = NULL,
       parallel::stopCluster(cl)
     }
   })
+  # clean up potential errors with warning
+  error_count <- 0
+  ranef_boot <- purrr::map(ranef_boot, \(x) {
+    if (inherits(x, "simpleError")) {
+      error_df <- data.frame(iter = NA_real_, 
+                             subject = NA_character_, 
+                             term = NA_character_, 
+                             effect = NA_real_)
+      error_count <- error_count + 1
+      return(error_df)
+    } else {
+      return(x)
+    }
+  })
+  if (error_count > 0) {
+    warning(paste0(error_count, " bootstrap summaries failed, likely due to computationally singular effects matrices. Take caution."))
+  }
   # summarize bootstrap resample
   lower_bound <- alpha / 2
   upper_bound <- 1 - alpha / 2
   ranef_sumy <- purrr::reduce(ranef_boot, rbind) %>%
+                as.data.frame() %>% 
+                na.omit() %>% 
                 dplyr::with_groups(c(subject, term),
                                    dplyr::summarise,
                                    QL = stats::quantile(effect, probs = lower_bound),
diff --git a/tests/testthat/test_scLANE.R b/tests/testthat/test_scLANE.R
index 5478051..7bfee2e 100644
--- a/tests/testthat/test_scLANE.R
+++ b/tests/testthat/test_scLANE.R
@@ -136,7 +136,7 @@ withr::with_output_sink(tempfile(), {
   # run GEE Wald test
   wald_test <- waldTestGEE(marge_mod_GEE_offset, mod.0 = null_mod_GEE)
   # run GLMM model -- no offset
-  glmm_mod <- fitGLMM(X_pred = pt_test,
+  glmm_mod <- fitGLMM(pt_test,
                       Y = counts_test[, 4],
                       id.vec = sim_data$subject,
                       adaptive = TRUE,
@@ -144,14 +144,14 @@ withr::with_output_sink(tempfile(), {
                       return.basis = TRUE,
                       return.GCV = TRUE)
   # run GLMM model -- with offset
-  glmm_mod_offset <- fitGLMM(X_pred = pt_test,
+  glmm_mod_offset <- fitGLMM(pt_test,
                              Y = counts_test[, 4],
                              Y.offset = cell_offset,
                              id.vec = sim_data$subject,
                              adaptive = TRUE,
                              M.glm = 3,
                              return.basis = TRUE,
-                             return.GCV = TRUE)
+                             return.GCV = FALSE)
   # bootstrap GLMM random effects
   re_sumy <- bootstrapRandomEffects(glmm_mod_offset,
                                     id.vec = sim_data$subject,
@@ -404,7 +404,7 @@ test_that("fitGLMM() output", {
   expect_equal(glmm_mod$model_type, "GLMM")
   expect_equal(glmm_mod_offset$model_type, "GLMM")
   expect_s3_class(re_sumy, "data.frame")
-  expect_equal(nrow(re_sumy), 6)
+  expect_equal(nrow(re_sumy), 4)
   expect_equal(ncol(re_sumy), 4)
 })