From 14f8f5d6d4f3fc536a676feadd2374849e24c548 Mon Sep 17 00:00:00 2001
From: Tanushree Tunstall <tanu@tunstall.in>
Date: Thu, 4 Aug 2022 19:48:09 +0100
Subject: [PATCH] generated lineage barplots and corr plots for conservation

---
 .../plotting_thesis/corr_plots_thesis.R       | 42 ++++++++++++++++++-
 .../lineage_barplots_combined.R               | 12 +++---
 .../plotting/plotting_thesis/preformatting.R  |  7 ++--
 3 files changed, 52 insertions(+), 9 deletions(-)

diff --git a/scripts/plotting/plotting_thesis/corr_plots_thesis.R b/scripts/plotting/plotting_thesis/corr_plots_thesis.R
index 9c02b4e..c282341 100644
--- a/scripts/plotting/plotting_thesis/corr_plots_thesis.R
+++ b/scripts/plotting/plotting_thesis/corr_plots_thesis.R
@@ -137,6 +137,46 @@ my_corr_pairs(corr_data_all = corr_df_ppi2
               , corr_value_size = 1)
 
 dev.off()
+
+# FIXME: ADD distance
 #==================
 # mCSSM-NA affinity
-#==================
\ No newline at end of file
+#==================
+
+
+
+
+# FIXME: ADD PROVEAN
+####################################################
+# CONSERVATION
+####################################################
+corr_df_cons = corr_plotdf[, corr_conservation_cols]
+
+color_coln = which(colnames(corr_df_cons) == "dst_mode")
+end = which(colnames(corr_df_cons) == drug)
+ncol_omit = 2 
+corr_end = end-ncol_omit
+
+#---------------------------
+# Output: Conservation corrP
+#----------------------------
+corr_consP =  paste0(outdir_images
+                     ,tolower(gene)
+                     ,"_corr_ppi2.svg" )
+
+cat("Corr plot conservation coloured dots:", corr_consP)
+svg(corr_consP, width = 10, height = 10)
+
+my_corr_pairs(corr_data_all = corr_df_cons
+              , corr_cols = colnames(corr_df_cons[1:corr_end])
+              , corr_method = "spearman" # other options: "pearson" or "kendall"
+              , colour_categ_col = colnames(corr_df_cons[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()
diff --git a/scripts/plotting/plotting_thesis/lineage_barplots_combined.R b/scripts/plotting/plotting_thesis/lineage_barplots_combined.R
index 01f6428..c9defaf 100644
--- a/scripts/plotting/plotting_thesis/lineage_barplots_combined.R
+++ b/scripts/plotting/plotting_thesis/lineage_barplots_combined.R
@@ -1,5 +1,6 @@
 # get plotting_dfs()
 # using dfs [lf and wf] from lineage_dfL
+# FIXME: add SNP diversity as the title
 #===============================
 # lineage sample and SNP count
 #===============================
@@ -54,15 +55,16 @@ lin_diversityP = lin_count_bp_diversity(lf_data = lineage_dfL[['lin_wf']]
 #=============================================
 lineage_bp_CL =  paste0(outdir_images
                         ,tolower(gene)
-                        ,"_lineage_bp_CL.svg")
+                        ,"_lineage_bp_CL_Tall.svg")
 
 cat("Lineage barplots:", lineage_bp_CL)
 svg(lineage_bp_CL, width = 8, height = 15)
 
 cowplot::plot_grid(lin_countP, lin_diversityP
-                                         #, labels = c("(a)", "(b)", "(c)", "(d)")
-                                         , nrow = 2
-                                         , labels = "AUTO"
-                                         , label_size = 25)
+                     #, labels = c("(a)", "(b)", "(c)", "(d)")
+                      , nrow = 2
+                     # , ncols = 2
+                      , labels = "AUTO"
+                      , label_size = 25)
 
 dev.off()
\ No newline at end of file
diff --git a/scripts/plotting/plotting_thesis/preformatting.R b/scripts/plotting/plotting_thesis/preformatting.R
index 74fbbf7..37e7cc0 100644
--- a/scripts/plotting/plotting_thesis/preformatting.R
+++ b/scripts/plotting/plotting_thesis/preformatting.R
@@ -235,11 +235,12 @@ corr_ppi2_colnames = c("mCSM-PPI2"
                        , "dst_mode"
                        , drug)
 
-
-corr_conservation = c("Consurf"
+#FIXME: Add provean
+corr_conservation_cols = c("Consurf"
+                      , "SNAP2"
                        , "MAF"             
                        , "Log (OR)"       
                        , "-Log (P)"
-                       , "interface_dist" 
                        , "dst_mode"
                        , drug)
+