From 5f10ad807526ce1bb5b9e146f94542710b2fc321 Mon Sep 17 00:00:00 2001
From: Tanushree Tunstall <tanu@tunstall.in>
Date: Wed, 23 Sep 2020 11:20:22 +0100
Subject: [PATCH] added fold and duet agreement to extreme_muts.R

---
 scripts/plotting/extreme_muts.R  | 23 +++++++++++++++++++++--
 scripts/plotting/plotting_data.R |  8 ++++----
 2 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/scripts/plotting/extreme_muts.R b/scripts/plotting/extreme_muts.R
index 65943aa..053fde9 100644
--- a/scripts/plotting/extreme_muts.R
+++ b/scripts/plotting/extreme_muts.R
@@ -10,6 +10,7 @@ setwd("~/git/LSHTM_analysis/scripts/plotting")
 getwd()
 
 #source("Header_TT.R")
+library(tidyverse)
 library(ggplot2)
 library(data.table)
 library(dplyr)
@@ -19,7 +20,6 @@ library(dplyr)
 #=========
 #source("combining_dfs_plotting.R")
 
-# FIXME: add a separate script to add foldx values and others
 source("output_tables.R")
 rm(df, merged_df3_short, df_output)
 
@@ -146,4 +146,23 @@ n_mult_muts_sites = sum(table(foo$n)) - (table(foo$n)[[1]] - table(foo$n)[[2]])
 cat("No of budding hotspots (sites with 2 mutations):", n_budding_sites
     , "\nNo. of sites with mutiple (>2) mutations:", n_mult_muts_sites)
 
-#==========================================================================
\ No newline at end of file
+#==========================================================================
+
+#==============================
+# agreement of foldx and DUET
+#==============================
+mcsm_foldx = merged_df3[which(merged_df3$duet_outcome != merged_df3$foldx_outcome),]
+
+mcsm_foldx$sign_comp = ifelse(sign(mcsm_foldx$duet_scaled)==sign(mcsm_foldx$ddg), "PASS", "FAIL")
+table(mcsm_foldx$sign_comp)
+
+# another way of checking
+merged_df3$sign_comp = ifelse(sign(merged_df3$duet_scaled)==sign(merged_df3$ddg), "PASS", "FAIL")
+table(merged_df3$sign_comp)
+
+disagreement = table(merged_df3$sign_comp)[2]/nrow(merged_df3)*100
+agreement = 100 - disagreement
+
+cat("There is", agreement, "% between mcsm and foldx predictions")
+
+##############################################################################
diff --git a/scripts/plotting/plotting_data.R b/scripts/plotting/plotting_data.R
index 0c15a78..2df3ef4 100755
--- a/scripts/plotting/plotting_data.R
+++ b/scripts/plotting/plotting_data.R
@@ -88,10 +88,10 @@ my_min = min(my_df$foldx_scaled); my_min
 my_max = max(my_df$foldx_scaled); my_max
 
 if (my_min == -1 && my_max == 1){
-  cat("PASS: foldx ddg successfully scaled b/w -1 and 1"
+  cat("\nPASS: foldx ddg successfully scaled b/w -1 and 1"
       , "\nProceeding with assigning foldx outcome category")
 }else{
-  cat("FAIL: could not scale foldx ddg values"
+  cat("\nFAIL: could not scale foldx ddg values"
       , "Aborting!")
 }
 
@@ -104,9 +104,9 @@ my_df$foldx_outcome = ifelse(my_df$ddg < 0, "Stabilising", "Destabilising")
 c2 = table(my_df$ddg < 0)
 
 if ( all(c1 == c2) ){
-  cat("PASS: foldx outcome successfully created")
+  cat("\nPASS: foldx outcome successfully created")
 }else{
-  cat("FAIL: foldx outcome could not be created. Aborting!")
+  cat("\nFAIL: foldx outcome could not be created. Aborting!")
   exit()
 }