From c8e21b928ca27eac0fa20c481bd057acb0fc05fc Mon Sep 17 00:00:00 2001
From: Tanushree Tunstall <tanu@tunstall.in>
Date: Fri, 27 Aug 2021 14:05:00 +0100
Subject: [PATCH] added functions for bp with stat and tested them

---
 scripts/functions/lf_bp_stability.R        | 100 ---------------------
 scripts/functions/lf_bp_with_stats.R       |  68 ++++++++++++++
 scripts/functions/test_lf_bp_with_stats.R  |  28 ++++++
 scripts/functions/test_lf_unpaired_stats.R |  12 +--
 4 files changed, 103 insertions(+), 105 deletions(-)
 delete mode 100644 scripts/functions/lf_bp_stability.R
 create mode 100644 scripts/functions/lf_bp_with_stats.R
 create mode 100644 scripts/functions/test_lf_bp_with_stats.R

diff --git a/scripts/functions/lf_bp_stability.R b/scripts/functions/lf_bp_stability.R
deleted file mode 100644
index 8ca4a7f..0000000
--- a/scripts/functions/lf_bp_stability.R
+++ /dev/null
@@ -1,100 +0,0 @@
-library(ggpubr)
-###################################################################
-
-my_unpaired_stats <- function(lf_data
-                              , lf_stat_value = "param_value"
-                              , lf_stat_group = "mutation_info"
-                              , lf_col_statvars = "param_type"
-                              , my_paired = FALSE
-                              , stat_adj = "none"){
-  
-  stat_formula = as.formula(paste0(lf_stat_value, "~", lf_stat_group))
-  
-  my_stat_df = compare_means(stat_formula
-                             , group.by = lf_col_statvars
-                             , data = lf_data
-                             , paired = my_paired
-                             , p.adjust.method =  stat_adj)  
-  
-  
-  return(my_stat_df)
-}
-  
-#####################
-# call stat function
-#####################
-stat_results_df <- my_unpaired_stats(lf_data =  lf_duet
-                  , lf_stat_value = "param_value"
-                  , lf_stat_group = "mutation_info"
-                  , lf_col_statvars = "param_type"
-                  , my_paired = FALSE
-                  , stat_adj = "none"
-)
-
-y_value = "param_value"
-
-#################################
-my_comparisons <- list( c("DM", "OM") )
-
-my_ats = 22 # axis text size
-my_als = 20 # axis label size
-my_fls = 20 # facet label size
-my_pts = 22 # plot title size
-
-####################################
-
-
-stat_bp_mut <- function(df
-                        , x_bp_cateog = "mutation_info"
-                        , y_var = "param_value"
-                        , facet_var = "param_type"
-                        , scales = "free_y"
-                        , title = ""
-                        , col_categ = "duet_outcome"
-                        , grp_comp = "my_comparisons"
-                        , stat_method = "wilcox.test"
-                        , my_paired = FALSE
-                        #, stat_label = "p.format")
-                        , stat_label = "p.signif" )
-
-
-#############################
-y_value = "param_value"
-
-
-p1 = ggplot(lf_duet, aes(x = mutation_info
-                          , y = eval(parse(text = y_value)) ))  + 
-  facet_wrap(~ param_type
-             , nrow = 1
-             , scales = "free_y") + 
-  geom_boxplot(fill = "white", outlier.colour = NA
-               #, position = position_dodge(width = 0.9)
-               , width = 0.2) +
-  geom_point(position = position_jitterdodge(dodge.width=0.01)
-             , alpha = 0.5
-             , show.legend = FALSE
-             , aes(colour = factor(duet_outcome))) +
-  theme(axis.text.x = element_text(size = my_ats)
-        , axis.text.y = element_text(size = my_ats
-                                     , angle = 0
-                                     , hjust = 1
-                                     , vjust = 0)
-        , axis.title.x = element_text(size = my_ats)
-        , axis.title.y = element_text(size = my_ats)
-        , plot.title = element_text(size = my_pts , hjust = 0.5, colour = "black", face = "bold")
-        , strip.background = element_rect(fill = "khaki2")
-        , strip.text.x = element_text(size = my_fls, colour = "black")
-        , legend.title = element_text(color = "black", size = my_als)
-        , legend.text = element_text(size = my_ats)
-        , legend.direction = "vertical") +
-  labs(title = "DUET"
-       , x = ""
-       , y = "")+ 
-  stat_compare_means(comparisons = my_comparisons
-                     , method = "wilcox.test"
-                     , paired = FALSE
-                     #, label = "p.format")
-                     , label = "p.signif")
-
-p1
-
diff --git a/scripts/functions/lf_bp_with_stats.R b/scripts/functions/lf_bp_with_stats.R
new file mode 100644
index 0000000..336c270
--- /dev/null
+++ b/scripts/functions/lf_bp_with_stats.R
@@ -0,0 +1,68 @@
+library(ggpubr)
+###################################################################
+
+####################################
+lf_bp_with_stats <- function(lf_df
+                        , x_grp = "mutation_info"
+                        , y_var = "param_value"
+                        , facet_var = "param_type"
+                        , n_facet_row = 1
+                        , y_scales = "free_y"
+                        , p_title = ""
+                        , colour_categ = ""
+                        , colour_bp_strip = "khaki2"
+                        , stat_grp_comp = c("DM", "OM")
+                        , stat_method = "wilcox.test"
+                        , my_paired = FALSE
+                        #, stat_label = "p.format")
+                        , stat_label = c("p.format", "p.signif")
+                        , my_ats = 22 # axis text size
+                        , my_als = 20 # axis label size
+                        , my_fls = 20 # facet label size
+                        , my_pts = 22 # plot title size
+) {
+  my_comparisonsL <- list( stat_grp_comp )
+
+  bp_statP <- ggplot(lf_df, aes(x = eval(parse(text = x_grp))
+                    , y = eval(parse(text = y_var)) ))  + 
+    
+    facet_wrap(~ eval(parse(text = facet_var))
+               , nrow = n_facet_row
+               , scales = y_scales) +
+    
+    geom_boxplot(fill = "white", outlier.colour = NA
+                 #, position = position_dodge(width = 0.9)
+                 , width = 0.2) +
+    
+    geom_point(position = position_jitterdodge(dodge.width = 0.01)
+               , alpha = 0.5
+               , show.legend = FALSE
+               , aes(colour = factor(eval(parse(text = colour_categ))) )) +
+    
+    theme(axis.text.x = element_text(size = my_ats)
+          , axis.text.y = element_text(size = my_ats
+                                       , angle = 0
+                                       , hjust = 1
+                                       , vjust = 0)
+          , axis.title.x = element_text(size = my_ats)
+          , axis.title.y = element_text(size = my_ats)
+          , plot.title = element_text(size = my_pts , hjust = 0.5, colour = "black", face = "bold")
+          , strip.background = element_rect(fill = colour_bp_strip)
+          , strip.text.x = element_text(size = my_fls, colour = "black")
+          , legend.title = element_text(color = "black", size = my_als)
+          , legend.text = element_text(size = my_ats)
+          , legend.direction = "vertical") +
+    
+    labs(title = p_title
+         , x = ""
+         , y = "")+ 
+  
+    stat_compare_means(comparisons = my_comparisonsL
+                       , method = stat_method
+                       , paired = my_paired
+                       #, label = "p.format")
+                       , label = stat_label[1])
+  
+  return(bp_statP)
+
+}
diff --git a/scripts/functions/test_lf_bp_with_stats.R b/scripts/functions/test_lf_bp_with_stats.R
new file mode 100644
index 0000000..4cfff9d
--- /dev/null
+++ b/scripts/functions/test_lf_bp_with_stats.R
@@ -0,0 +1,28 @@
+setwd("~/git/LSHTM_analysis/scripts/plotting/")
+
+source("../functions/lf_bp_with_stats.R")
+
+######################
+# call function
+######################
+# Note: Data
+# run other_plots_data.R
+# to get the long format data to test this function 
+
+lf_bp_with_stats(lf_df = lf_dynamut2
+                       , x_grp = "mutation_info"
+                       , y_var = "param_value"
+                       , facet_var = "param_type"
+                       , n_facet_row = 1
+                       , y_scales = "free_y"
+                       , p_title = "Dynamut2"
+                       , colour_categ = "ddg_dynamut2_outcome"
+                       , stat_grp_comp = c("DM", "OM")
+                       , stat_method = "wilcox.test"
+                       , my_paired = FALSE
+                       #, stat_label = "p.format")
+                       , stat_label = c("p.format", "p.signif")
+                       , my_ats = 22 # axis text size
+                       , my_als = 20 # axis label size
+                       , my_fls = 20 # facet label size
+                       , my_pts = 22 )# plot title size 
diff --git a/scripts/functions/test_lf_unpaired_stats.R b/scripts/functions/test_lf_unpaired_stats.R
index 9ec4aac..a1ff9d1 100644
--- a/scripts/functions/test_lf_unpaired_stats.R
+++ b/scripts/functions/test_lf_unpaired_stats.R
@@ -1,13 +1,15 @@
 setwd("~/git/LSHTM_analysis/scripts/functions")
 source("lf_unpaired_stats.R")
 
+#####################
+# call stat function()
+# a useful way to check stats
+# for any lf data
+#####################
+# Note: Data
 # run other_plots_data.R
-# to get the df you want to test this function 
+# to get the long format data to test this function 
 
-
-#####################
-# call stat function
-#####################
 stat_results_df <- lf_unpaired_stats(lf_data =  lf_duet
                   , lf_stat_value = "param_value"
                   , lf_stat_group = "mutation_info"