From de599ed5b8b9ec1d8450ab5cf85a32b2cd303dbb Mon Sep 17 00:00:00 2001
From: Bryan Roessler <bryanroessler@gmail.com>
Date: Thu, 5 Sep 2024 20:33:30 -0400
Subject: [PATCH] Use a single N value

---
 .../apps/r/calculate_interaction_zscores5.R   | 28 +++++++++++--------
 1 file changed, 16 insertions(+), 12 deletions(-)

diff --git a/workflow/apps/r/calculate_interaction_zscores5.R b/workflow/apps/r/calculate_interaction_zscores5.R
index 718815af..26d239cc 100644
--- a/workflow/apps/r/calculate_interaction_zscores5.R
+++ b/workflow/apps/r/calculate_interaction_zscores5.R
@@ -183,20 +183,23 @@ process_strains <- function(df) {
 
 # Calculate summary statistics for all variables
 calculate_summary_stats <- function(df, variables, group_vars = c("conc_num", "conc_num_factor")) {
+  # Calculate summary statistics, including a single N based on L
   summary_stats <- df %>%
     group_by(across(all_of(group_vars))) %>%
-    reframe(across(all_of(variables), list(
-      mean = ~mean(., na.rm = TRUE),
-      median = ~median(., na.rm = TRUE),
-      max = ~ifelse(all(is.na(.)), NA, max(., na.rm = TRUE)),
-      min = ~ifelse(all(is.na(.)), NA, min(., na.rm = TRUE)),
-      sd = ~sd(., na.rm = TRUE),
-      se = ~ifelse(sum(!is.na(.)) > 1, sd(., na.rm = TRUE) / sqrt(sum(!is.na(.)) - 1), NA)
-      # TODO: not in original stats but better to do here than in calculate_interactions?
-      # z_max = ~ifelse(sd(., na.rm = TRUE) == 0 | all(is.na(.)), NA,
-      #   (max(., na.rm = TRUE) - mean(., na.rm = TRUE)) / sd(., na.rm = TRUE))  # Z-score
-    ), .names = "{.fn}_{.col}")) %>%
-    mutate(N = sum(!is.na(L))) # count of non-NA L values
+    reframe(
+      N = sum(!is.na(L)),  # Single N based on L
+      across(all_of(variables), list(
+        mean = ~mean(., na.rm = TRUE),
+        median = ~median(., na.rm = TRUE),
+        max = ~ifelse(all(is.na(.)), NA, max(., na.rm = TRUE)),
+        min = ~ifelse(all(is.na(.)), NA, min(., na.rm = TRUE)),
+        sd = ~sd(., na.rm = TRUE),
+        se = ~ifelse(sum(!is.na(.)) > 1, sd(., na.rm = TRUE) / sqrt(sum(!is.na(.)) - 1), NA)
+        # TODO: not in original stats but better to do here than in calculate_interactions?
+        # z_max = ~ifelse(sd(., na.rm = TRUE) == 0 | all(is.na(.)), NA,
+        #   (max(., na.rm = TRUE) - mean(., na.rm = TRUE)) / sd(., na.rm = TRUE))  # Z-score
+      ), .names = "{.fn}_{.col}")
+    )
 
   # Join the summary stats back to the original dataframe
   df_with_stats <- left_join(df, summary_stats, by = group_vars)
@@ -205,6 +208,7 @@ calculate_summary_stats <- function(df, variables, group_vars = c("conc_num", "c
   return(list(summary_stats = summary_stats, df_with_stats = df_with_stats))
 }
 
+
 # Calculate interaction scores
 calculate_interaction_scores <- function(df, max_conc, variables, group_vars = c("OrfRep", "Gene", "num")) {