added foldx scaled and foldx outcome to plotting_data.R

scripts/plotting/plotting_data.R

@@ -63,6 +63,52 @@ my_df = read.csv(infile_params, header = T)
 
 cat("\nInput dimensions:", dim(my_df)) 
 
+###########################
+# add foldx outcome category
+# and foldx scaled values 
+
+# This will enable to always have these variables available
+# when calling for plots
+###########################
+
+#------------------------------
+# adding foldx scaled values
+# scale data b/w -1 and 1
+#------------------------------
+n = which(colnames(my_df) == "ddg"); n 
+
+my_min = min(my_df[,n]); my_min 
+my_max = max(my_df[,n]); my_max 
+
+my_df$foldx_scaled = ifelse(my_df[,n] < 0
+                            , my_df[,n]/abs(my_min)
+                            , my_df[,n]/my_max) 
+# sanity check
+my_min = min(my_df$foldx_scaled); my_min 
+my_max = max(my_df$foldx_scaled); my_max
+
+if (my_min == -1 && my_max == 1){
+  cat("PASS: foldx ddg successfully scaled b/w -1 and 1"
+      , "\nProceeding with assigning foldx outcome category")
+}else{
+  cat("FAIL: could not scale foldx ddg values"
+      , "Aborting!")
+}
+
+#------------------------------
+# adding foldx outcome category
+# ddg<0 = "Stabilising" (-ve)
+#------------------------------
+c1 = table(my_df$ddg < 0)
+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")
+}else{
+  cat("FAIL: foldx outcome could not be created. Aborting!")
+  exit()
+}
 
 ###########################
 # extract unique mutation entries