aadded complete obs count for corr plots and added placeholder for mcsm-na

scripts/plotting/plotting_thesis/corr_plots_thesis.R

@@ -41,6 +41,8 @@ corr_ps_colnames = c("DUET"
 
 corr_ps_colnames%in%colnames(corr_plotdf)
 corr_df_ps = corr_plotdf[, corr_ps_colnames]
+complete_obs_ps = nrow(corr_df_ps) - sum(is.na(corr_df_ps$`Log(OR)`))
+cat("\nComplete muts for Conservation for", gene, ":", complete_obs_ps)
 
 color_coln = which(colnames(corr_df_ps) == "dst_mode")
 end = which(colnames(corr_df_ps) == drug)
@@ -91,6 +93,8 @@ corr_lig_colnames = c("mCSM-lig"
 corr_lig_colnames%in%colnames(corr_plotdf)
 corr_df_lig = corr_plotdf[, corr_lig_colnames]
 corr_df_lig = corr_df_lig[corr_df_lig["Lig-Dist"]<DistCutOff,]
+complete_obs_lig = nrow(corr_df_lig) - sum(is.na(corr_df_lig$`Log(OR)`))
+cat("\nComplete muts for lig affinity for", gene, ":", complete_obs_lig)
 
 color_coln = which(colnames(corr_df_lig) == "dst_mode")
 end = which(colnames(corr_df_lig) == drug)
@@ -134,6 +138,8 @@ corr_ppi2_colnames = c("mCSM-PPI2"
 corr_ppi2_colnames%in%colnames(corr_plotdf)
 corr_df_ppi2 = corr_plotdf[, corr_ppi2_colnames]
 corr_df_ppi2 = corr_df_ppi2[corr_df_ppi2["PPI-Dist"]<DistCutOff,]
+complete_obs_ppi2 = nrow(corr_df_ppi2) - sum(is.na(corr_df_ppi2$`Log(OR)`))
+cat("\nComplete muts for ppi2 affinity for", gene, ":", complete_obs_ppi2)
 
 color_coln = which(colnames(corr_df_ppi2) == "dst_mode")
 end = which(colnames(corr_df_ppi2) == drug)
@@ -168,9 +174,51 @@ dev.off()
 #==================
 # mCSSM-NA affinity
 #==================
+#================
+# NA affinity
+#================
+corr_na_colnames = c("mCSM-NA"
+                     , "MAF"             
+                     , "Log(OR)"       
+                     , "-Log(P)"
+                     , "NA-Dist"  # "interface_dist"
+                     , "dst_mode"
+                     , drug)
 
+corr_na_colnames%in%colnames(corr_plotdf)
+corr_df_na = corr_plotdf[, corr_na_colnames]
+corr_df_na = corr_df_na[corr_df_na["NA-Dist"]<DistCutOff,]
+complete_obs_na = nrow(corr_df_na) - sum(is.na(corr_df_na$`Log(OR)`))
+cat("\nComplete muts for NA affinity for", gene, ":", complete_obs_na)
 
+color_coln = which(colnames(corr_df_na) == "dst_mode")
+end = which(colnames(corr_df_na) == drug)
+ncol_omit = 3 #omit dist col
+corr_end = end-ncol_omit
 
+#------------------------
+# Output: mCSM-NA corrP
+#------------------------
+corr_naP =  paste0(outdir_images
+                   ,tolower(gene)
+                   ,"_corr_na.svg" )
+
+cat("Corr plot mCSM-NA with coloured dots:", corr_naP)
+svg(corr_naP, width = 10, height = 10)
+
+my_corr_pairs(corr_data_all = corr_df_na
+              , corr_cols = colnames(corr_df_na[1:corr_end])
+              , corr_method = "spearman" 
+              , colour_categ_col = colnames(corr_df_na[color_coln]) #"dst_mode"
+              , categ_colour =  c("red", "blue")
+              , density_show = F
+              , hist_col = "coral4"
+              , dot_size = 2
+              , ats = 1.5
+              , corr_lab_size = 3
+              , corr_value_size = 1)
+
+dev.off()
 
 ####################################################
 # CONSERVATION
@@ -183,11 +231,11 @@ corr_conservation_cols = c("ConSurf"
                            , "dst_mode"
                            , drug)
 
-####################################################
 colnames(corr_plotdf)
 corr_conservation_cols%in%colnames(corr_plotdf)
-
 corr_df_cons = corr_plotdf[, corr_conservation_cols]
+complete_obs_cons = nrow(corr_df_cons) - sum(is.na(corr_df_cons$`Log(OR)`))
+cat("\nComplete muts for Conservation for", gene, ":", complete_obs_cons)
 
 color_coln = which(colnames(corr_df_cons) == "dst_mode")
 end = which(colnames(corr_df_cons) == drug)