From eb1c04bbaba6c8c30c14259311ba961558d4984a Mon Sep 17 00:00:00 2001
From: Bryan Roessler <bryanroessler@gmail.com>
Date: Tue, 17 Sep 2024 16:56:12 -0400
Subject: [PATCH] Fix rank vars names

---
 .../apps/r/calculate_interaction_zscores.R    | 48 +++++++++++--------
 1 file changed, 28 insertions(+), 20 deletions(-)

diff --git a/qhtcp-workflow/apps/r/calculate_interaction_zscores.R b/qhtcp-workflow/apps/r/calculate_interaction_zscores.R
index abd19b81..ec7763ce 100644
--- a/qhtcp-workflow/apps/r/calculate_interaction_zscores.R
+++ b/qhtcp-workflow/apps/r/calculate_interaction_zscores.R
@@ -727,7 +727,7 @@ generate_interaction_plot_configs <- function(df, variables) {
     # Box plot config
     configs[[length(configs) + 1]] <- modifyList(plot_settings, list(
       plot_type = "box",
-      title = sprintf("%s      %s (Boxplot)", df_filtered$OrfRep[1], df_filtered$Gene[1]),
+      title = sprintf("%s      %s (box plot)", df_filtered$OrfRep[1], df_filtered$Gene[1]),
       error_bar = FALSE
     ))
   }
@@ -746,11 +746,11 @@ generate_rank_plot_configs <- function(df_filtered, variables, is_lm = FALSE) {
     for (sd_band in sd_bands) {
       # Determine columns based on whether it's lm or not
       if (is_lm) {
-        rank_var <- paste0(variable, "_Rank_lm")
+        rank_var <- paste0("Rank_lm_", variable)
         zscore_var <- paste0("Z_lm_", variable)
         y_label <- paste("Int Z score", variable)
       } else {
-        rank_var <- paste0(variable, "_Rank")
+        rank_var <- paste0("Rank_", variable)
         zscore_var <- paste0("Avg_Zscore_", variable)
         y_label <- paste("Avg Z score", variable)
       }
@@ -819,19 +819,27 @@ generate_rank_plot_configs <- function(df_filtered, variables, is_lm = FALSE) {
           )
         )
       } else {
-        x_var <- paste0(variable, "_Rank")
-        y_var <- paste0(variable, "_Rank_lm")
+        x_var <- paste0("Rank", variable)
+        y_var <- paste0("Rank_lm_", variable)
         title_suffix <- paste("Rank Avg Zscore vs lm", variable)
         rectangles <- NULL
       }
+
+      print(df_filtered %>% select(all_of(c(x_var, y_var))))
+
+      # Check if there is sufficient variation
+      if (length(unique(df_filtered[[x_var]])) < 2 || length(unique(df_filtered[[y_var]])) < 2) {
+        message("Not enough variation in ", x_var, " or ", y_var, " to fit linear model. Skipping.")
+        next
+      }
       
       # Fit linear model
       lm_fit <- lm(df_filtered[[y_var]] ~ df_filtered[[x_var]], data = df_filtered)
       
-      # Check for perfect fit
-      if (summary(lm_fit)$sigma == 0) {
-        next  # Skip this iteration if the fit is perfect
-      }
+      # # Check for perfect fit
+      # if (summary(lm_fit)$sigma == 0) {
+      #   next  # Skip this iteration if the fit is perfect
+      # }
       
       r_squared <- summary(lm_fit)$r.squared
       
@@ -1158,7 +1166,7 @@ main <- function() {
       df_after = df_na_filtered_stats,
     )
 
-    plate_analysis_box_plot_configs <- generate_plate_analysis_plot_configs(
+    plate_analysis_boxplot_configs <- generate_plate_analysis_plot_configs(
       variables = summary_vars,
       df_before = df_filtered_stats,
       df_after = df_na_filtered_stats,
@@ -1206,16 +1214,16 @@ main <- function() {
       )
     )
 
-    message("Generating quality control plots")
-    generate_and_save_plots(out_dir_qc, "L_vs_K_before_quality_control", l_vs_k_plots)
-    generate_and_save_plots(out_dir_qc, "frequency_delta_background", frequency_delta_bg_plots)
-    generate_and_save_plots(out_dir_qc, "L_vs_K_above_threshold", above_threshold_plots)
-    generate_and_save_plots(out_dir_qc, "plate_analysis", plate_analysis_plot_configs)
-    generate_and_save_plots(out_dir_qc, "plate_analysis_boxplots", plate_analysis_boxplot_configs)
-    generate_and_save_plots(out_dir_qc, "plate_analysis_no_zeros", plate_analysis_no_zeros_plot_configs)
-    generate_and_save_plots(out_dir_qc, "plate_analysis_no_zeros_boxplots", plate_analysis_no_zeros_boxplot_configs)
-    generate_and_save_plots(out_dir_qc, "L_vs_K_for_strains_2SD_outside_mean_K", l_outside_2sd_k_plots)
-    generate_and_save_plots(out_dir_qc, "delta_background_vs_K_for_strains_2sd_outside_mean_K", delta_bg_outside_2sd_k_plots)
+    # message("Generating quality control plots")
+    # generate_and_save_plots(out_dir_qc, "L_vs_K_before_quality_control", l_vs_k_plots)
+    # generate_and_save_plots(out_dir_qc, "frequency_delta_background", frequency_delta_bg_plots)
+    # generate_and_save_plots(out_dir_qc, "L_vs_K_above_threshold", above_threshold_plots)
+    # generate_and_save_plots(out_dir_qc, "plate_analysis", plate_analysis_plot_configs)
+    # generate_and_save_plots(out_dir_qc, "plate_analysis_boxplots", plate_analysis_boxplot_configs)
+    # generate_and_save_plots(out_dir_qc, "plate_analysis_no_zeros", plate_analysis_no_zeros_plot_configs)
+    # generate_and_save_plots(out_dir_qc, "plate_analysis_no_zeros_boxplots", plate_analysis_no_zeros_boxplot_configs)
+    # generate_and_save_plots(out_dir_qc, "L_vs_K_for_strains_2SD_outside_mean_K", l_outside_2sd_k_plots)
+    # generate_and_save_plots(out_dir_qc, "delta_background_vs_K_for_strains_2sd_outside_mean_K", delta_bg_outside_2sd_k_plots)
 
     # Process background strains
     bg_strains <- c("YDL227C")