From ebac3daf14161981995520818d552a8c894faa21 Mon Sep 17 00:00:00 2001
From: Emmanouil Sifakis <emmanouil.sifakis@ki.se>
Date: Thu, 19 Feb 2026 18:27:35 +0100
Subject: [PATCH] Fix AUTO NA handling in ER/HER2 subgroup counts (na.rm=TRUE)
 (closes #127)

---
 NEWS.md       |  5 ++---
 R/Utilities.R | 27 +++++++++++++++++++--------
 2 files changed, 21 insertions(+), 11 deletions(-)

diff --git a/NEWS.md b/NEWS.md
index 146ff5c..9f9df9e 100755
--- a/NEWS.md
+++ b/NEWS.md
@@ -25,15 +25,14 @@
     -   Readiness uses the same minimums used by AUTO (sourced programmatically; no duplicated thresholds).
 -   **Shorter notifications.**\
     -Routine toasts (e.g., "Step 1 complete. Proceed to Step 2.") now auto-dismiss sooner to reduce UI clutter.
--   **Phenodata normalization (Mapping):** Accepts flexible ER/HER2/TN encodings and
-normalizes to canonical forms (`ER+/ER-`, `HER2+/HER2-`, `TN/nonTN`). 
-Ambiguous `HER2="2+"` remains as-is and raises a warning.
+-   **Phenodata normalization (Mapping):** Accepts flexible ER/HER2/TN encodings and normalizes to canonical forms (`ER+/ER-`, `HER2+/HER2-`, `TN/nonTN`). Ambiguous `HER2="2+"` remains as-is and raises a warning.
 
 ### Bug fixes
 
 -   **TN cohorts + ssBC**: `BS_Multi()` now respects TN cohorts when methods are specified manually;
 `ssBC`/`ssBC.v2` switch to `s = "TN"` / `"TN.v2"` when a `TN` column indicates a TN cohort.
 Falls back to `s = "ER"` / `"ER.v2"` otherwise.
+-   **AUTO**: Fixed a crash in BS_Multi(methods="AUTO") when ER and/or HER2 contained missing values (NA).
 -   **AUTO internals**: fixed variable name typo (`samples_ERHER2.icd`).
 -   **Mapping():** Robust ENTREZID coercion (from `as.character()` to `as.integer()` with suppressed warnings).
 -   **cIHC.itr**: `outList$distances` now returned as numeric matrix.
diff --git a/R/Utilities.R b/R/Utilities.R
index ad21399..e5bf9a5 100755
--- a/R/Utilities.R
+++ b/R/Utilities.R
@@ -342,16 +342,27 @@ get_methods <- function(pheno) {
     } else if (!all(c("ER", "HER2") %in% colnames(pheno))) {
         stop("The 'AUTO' mode requires both 'ER' and 'HER2' columns in the 'pheno' dataframe.")
     } else {
+      
+        pheno <- .normalize_er_her2_tn(pheno)
+        
+        # Detect missing receptor annotations (optional)
+        n_na_ER   <- sum(is.na(pheno$ER))
+        n_na_HER2 <- sum(is.na(pheno$HER2))
+        if (n_na_ER > 0L || n_na_HER2 > 0L) {
+          .msg("AUTO: missing ER/HER2 values detected (ER NA=%d, HER2 NA=%d). These samples are excluded from subgroup counts.",
+               n_na_ER, n_na_HER2, origin = "AUTO")
+        }
+      
         # Calculate sample sizes
         sample_counts <- with(pheno, table(ER, HER2)) # (kept for future debugging)
-        n_ERpos <- sum(pheno$ER == "ER+")
-        n_ERneg <- sum(pheno$ER == "ER-")
-
-        n_ERnegHER2pos <- sum(pheno$ER == "ER-" & pheno$HER2 == "HER2+")
-        n_ERnegHER2neg <- sum(pheno$ER == "ER-" & pheno$HER2 == "HER2-")
-        n_ERposHER2pos <- sum(pheno$ER == "ER+" & pheno$HER2 == "HER2+")
-        n_ERposHER2neg <- sum(pheno$ER == "ER+" & pheno$HER2 == "HER2-")
-
+        n_ERpos <- sum(pheno$ER == "ER+", na.rm = TRUE)
+        n_ERneg <- sum(pheno$ER == "ER-", na.rm = TRUE)
+        
+        n_ERnegHER2pos <- sum(pheno$ER == "ER-" & pheno$HER2 == "HER2+", na.rm = TRUE)
+        n_ERnegHER2neg <- sum(pheno$ER == "ER-" & pheno$HER2 == "HER2-", na.rm = TRUE)
+        n_ERposHER2pos <- sum(pheno$ER == "ER+" & pheno$HER2 == "HER2+", na.rm = TRUE)
+        n_ERposHER2neg <- sum(pheno$ER == "ER+" & pheno$HER2 == "HER2-", na.rm = TRUE)
+        
         # Set thresholds
         n_ERpos_threshold <- 15 # simulation-based cut-off
         n_ERneg_threshold <- 18 # simulation-based cut-off