From 9dba59a71e572f1450a1b7e9a65c055a9bdeea1c Mon Sep 17 00:00:00 2001
From: Daena Rys <rysdaena8@gmail.com>
Date: Tue, 1 Sep 2026 16:48:16 +0300
Subject: [PATCH] Clarify IntegratedLearner sample requirements

---
 .../extra_material/integrated_learner.qmd     | 24 +++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/inst/pages/extra_material/integrated_learner.qmd b/inst/pages/extra_material/integrated_learner.qmd
index 2f70f13f..5381577a 100644
--- a/inst/pages/extra_material/integrated_learner.qmd
+++ b/inst/pages/extra_material/integrated_learner.qmd
@@ -54,6 +54,30 @@ mae <- MultiAssayExperiment(
 mae
 ```
 
+`IntegratedLearnerFromMAE()` requires complete cases across the selected
+experiments. Every experiment must contain the same samples, and the sample
+names must match character-for-character. A `sampleMap` records relationships
+between samples and subjects in a `MultiAssayExperiment`, but it does not
+impute missing measurements or harmonize names for IntegratedLearner.
+
+Check this requirement before fitting the model:
+
+```{r}
+#| label: check_sample_names
+
+sample_ids <- lapply(experiments(mae), colnames)
+stopifnot(all(vapply(
+    sample_ids,
+    identical,
+    logical(1),
+    y = sample_ids[[1]]
+)))
+```
+
+For other datasets, rename columns to a shared sample or patient identifier
+and subset every experiment to the complete-case intersection before creating
+the `MultiAssayExperiment`.
+
 Let's first check how many patients are in each group.
 
 ```{r}