fix many plot functions to stop them using the "g=ggplot()" pattern,

which annoyingly throws away lots of useful data that RShiny needs for
clickable plots. Also split the "flame bar" for ligand distance out into
separate functions in generate_distance_colour_map.R. This can now be
easily incorporated into any "wide" graph showing all positions.

scripts/functions/bp_subcolours.R

@@ -2,6 +2,7 @@
 # 1b: Define function: coloured barplot by subgroup
 # LINK: https://stackoverflow.com/questions/49818271/stacked-barplot-with-colour-gradients-for-each-bar
 #########################################################
+source("~/git/LSHTM_analysis/scripts/functions/generate_distance_colour_map.R")
 
 ColourPalleteMulti = function(df, group, subgroup){
   
@@ -35,6 +36,7 @@ ColourPalleteMulti = function(df, group, subgroup){
 
 bp_stability_hmap <- function(plotdf = merged_df3
                               , xvar_colname = "position"
+                              , yvar_colname = 'duet_scaled' #FIXME: temp, remove
                               #, bar_col_colname = "group"
                               , stability_colname = "duet_scaled"
                               , stability_outcome_colname = "duet_outcome"
@@ -46,8 +48,27 @@ bp_stability_hmap <- function(plotdf = merged_df3
                               , my_pts  = 20  # plot-title size
                               , my_xlab = "Position"
                               , my_ylab = "No. of nsSNPs"
+                              
+                              # Custom 2: x-axis: geom tiles ~ lig distance
+                              #, A_xvar_lig = T
+                              , lig_dist_colname = LigDist_colname # from globals
+                              , tpos0 = 0 # 0 is a magic number that does my sensible default
+                              , tW0 = 1
+                              , tH0 = 0.2
+                              
+                              
+                              
 )
 {
+  ################################################
+  # Custom 2: x-axis geom tiles ~ lig distance
+  ################################################
+  
+  #=========================
+  # Build data with colours
+  # ~ ligand distance
+  #=========================
+  plotdf = generate_distance_colour_map(plotdf, yvar_colname = stability_colname, debug=TRUE)
   
   # order the df by position and ensure it is a factor
   plotdf = plotdf[order(plotdf[[xvar_colname]]), ]
@@ -74,17 +95,15 @@ bp_stability_hmap <- function(plotdf = merged_df3
   
   cat("\nNo. of sub colours generated:", length(subcols_bp_hmap))
   
+  
   #-------------------------------
   # Generate the subcols barplot
   #-------------------------------
-  
+  cowplot::plot_grid(
-  #g = ggplot(plotdf, aes(x = factor(position, ordered = T)))
+    ggplot(plotdf, aes_string(x = xvar_colname
-  g = ggplot(plotdf, aes_string(x = xvar_colname
                               # , ordered = T)
-  ))
+    )) +
-  
+      geom_bar(aes(fill = group)
-  
-  OutWidePlot = g + geom_bar(aes(fill = group)
                , colour = "grey") +
       
       scale_fill_manual( values = subcols_bp_hmap
@@ -102,10 +121,19 @@ bp_stability_hmap <- function(plotdf = merged_df3
              , axis.title.y = element_text(size = my_yaxts )
              , plot.title = element_text(size = my_pts
                                          , hjust = 0.5)) +
+      geom_tile(aes(, tpos0 # heat-mapped distance tiles along the bot
+                    , width  = tW0
+                    , height = tH0)
+                , fill     = plotdf$ligD_colours
+                , colour   = plotdf$ligD_colours
+                , linetype = "blank") +
       
       labs(title = p_title
            , x = my_xlab
-         , y = my_ylab)
+           , y = my_ylab),
-  
+    generate_distance_legend(plotdf, yvar_colname = stability_colname),
-  return(OutWidePlot)
+    ncol = 2,
+    #align = "hv",
+    rel_widths = c(9/10, 0.4/10)
+  )
 }

scripts/functions/consurfP.R

@@ -239,29 +239,11 @@ wideP_consurf <- function(plotdf
   my_xlim = length(unique(plotdf[[xvar_colname]])); my_xlim
   ymin    = min(plotdf[[yvar_colname]]); ymin
   ymax    = max(plotdf[[yvar_colname]]); ymax
-  # DEBUG
-  # x_factor = as.factor(plotdf[[xvar_colname]])
-  # y_colour_factor = as.factor(plotdf[[yvar_colourN_colname]])
-  
-  # cat('XXXX THIS:', str(x_factor))
-  # 
-  # g = ggplot(plotdf, aes_string(x = x_factor)
-  #                            , y = yvar_colname
-  #                            , colour = y_colour_factor)
 
   g = ggplot(plotdf, aes_string(x = sprintf("factor(%s)", xvar_colname)
                                 , y = yvar_colname
                                 , colour = sprintf("factor(%s)", yvar_colourN_colname)
                                 ))
-  # g = ggplot(plotdf, aes_string(x = "position")
-  #                               , y = "consurf_score"
-  #                               , colour = "consurf_colour_rev") 
-  # g = ggplot(plotdf, aes_string(x = "position")
-  #                               , y = "consurf_score"
-  #                               , colour = "consurf_colour_rev") 
-  
-  # DEBUG
-  #g0=g
 
   "if SPECIAL do SPECIAL THING, otherwise do NORMAL THING"
   if (plot_type == "bar"){
@@ -275,8 +257,6 @@ wideP_consurf <- function(plotdf
                       , clip = "off") +
 geom_point(size = p_size) +
 scale_colour_manual(values   = point_colours)
-    #                     , labels = leg_labels
-    #                     , name   = leg_title1)
   }
 
   if (plot_error_bars){
@@ -424,29 +404,6 @@ wideP_consurf <- function(plotdf
 }else{
   out = g1
 }    
-  
-  #####################################################
-  # #============================================
-  # # x-axis: geom_tiles ~ ligand distance
-  # #============================================
-  # if(A_xvar_lig){
-  #   cat("\nColouring x-axis aa based on", lig_dist_colname
-  #       , "\nNo. of colours:", n_colours)
-  #   
-  #   g2 = g1 + geom_tile(aes(, tpos0
-  #                           , width  = tW0
-  #                           , height = tH0)
-  #                       , fill     = plotdf$ligD_colours
-  #                       , colour   = plotdf$ligD_colours
-  #                       , linetype = "blank")
-  #   
-  #   #cat("Nrows of plot df", length(plotdf$ligD_colours))
-  #   out = g2
-  # }else{
-  #   out = g1
-  # }
-  # 
-  ############################################################################################  
 #==============================================
 # x-axis: geom_tiles ~ active sites and others
 #==============================================

scripts/functions/generate_distance_colour_map.R

@@ -0,0 +1,121 @@
+# takes a dataframe and returns the same dataframe with two extra columns for colours and position
+generate_distance_colour_map = function(plotdf,
+                                        xvar_colname = "position",
+                                        yvar_colname = 'duet_scaled',
+                                        lig_dist_colname = "ligand_distance",
+                                        lig_dist_colours = c("green", "yellow", "orange", "red"),
+                                        #tpos0 = 0,
+                                        #tpos1 = 0,
+                                        #tpos2 = 0,
+                                        #tpos3 = 0,
+                                        debug = FALSE
+)
+{
+  #-------------------
+  # x and y axis
+  # range, scale, etc
+  #-------------------
+  my_xlim = length(unique(plotdf[[yvar_colname]])); my_xlim
+  ymin    = min(plotdf[[yvar_colname]]); ymin
+  ymax    = max(plotdf[[yvar_colname]]); ymax
+  
+  #if (tpos0 == 0){
+  #  tpos0 = ymin-0.5
+  #}
+  #if (tpos1 == 0){
+  #  tpos1 = ymin-0.65
+  #}
+  #if (tpos2 == 0){
+  #  tpos2 = ymin-0.75
+  #}
+  #if (tpos3 == 0){
+  #  tpos3 = ymin-0.85
+  #}
+  if (debug) {
+    cat("\nAnnotating x-axis ~", lig_dist_colname, "requested...")
+  }
+  #-------------------------------------
+  # round column values: to colour by
+  #--------------------------------------
+  #plotdf = plotdf[order(plotdf[[lig_dist_colname]]),]
+  plotdf['lig_distR'] = round(plotdf[[lig_dist_colname]], digits = 0)
+  #head(plotdf['lig_distR'])
+  
+  #-------------------------------------
+  # ligand distance range, min, max, etc
+  #--------------------------------------
+  lig_min  = min(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_min
+  lig_max  = max(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_max 
+  lig_mean = round(mean(round(plotdf[[lig_dist_colname]]), na.rm = T)); lig_mean
+  
+  #-------------------------------------
+  # Create mapping colour key
+  #--------------------------------------
+  # sorting removes NA, so that n_colours  ==  length(ligD_valsR)
+  n_colours  = length(sort(unique(round(plotdf[[lig_dist_colname]], digits = 0)))); n_colours
+  
+  lig_cols = colorRampPalette(lig_dist_colours)(n_colours); lig_cols
+  ligD_valsR = sort(unique(round(plotdf[[lig_dist_colname]], digits = 0))); ligD_valsR
+  length(ligD_valsR)
+  
+  if (debug) {
+    if (n_colours == length(ligD_valsR)) {
+      cat("\nStarting: mapping b/w"
+          , lig_dist_colname
+          , "and colours")
+    }else{
+      cat("\nCannot start mapping b/w", lig_dist_colname, "and colours..."
+          , "\nLength mismatch:"
+          , "No. of colours: ", n_colours
+          , "\nValues to map:", length(ligD_valsR))
+    }
+  }
+  
+  ligDcolKey <- data.frame(ligD_colours = lig_cols
+                           , lig_distR = ligD_valsR); ligDcolKey
+  names(ligDcolKey)
+  if (debug) {
+    cat("\nSuccessful: Mapping b/w", lig_dist_colname, "and colours")
+  }
+  #-------------------------------------
+  # merge colour key with plotdf
+  #-------------------------------------- 
+  plotdf = merge(plotdf, ligDcolKey, by = 'lig_distR')
+  
+  plotdf_check = as.data.frame(cbind(position    = plotdf[[xvar_colname]]
+                                     , ligD      = plotdf[[lig_dist_colname]]
+                                     , ligDR     = plotdf$lig_distR
+                                     , ligD_cols = plotdf$ligD_colours))
+  return(plotdf)
+}
+
+generate_distance_legend = function(plotdf,
+                                    yvar_colname,
+                                    xvar_colname = 'position',
+                                    lig_dist_colname = "ligand_distance",
+                                    legend_title = "Ligand\nDistance"
+                                    )
+{
+  # build legend for ligand distance "heat bar"
+  lig_min  = min(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_min
+  lig_max  = max(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_max 
+  lig_mean = round(mean(round(plotdf[[lig_dist_colname]]), na.rm = T)); lig_mean
+  
+  labels  = seq(lig_min, lig_max, len = 5); labels
+  labelsD = round(labels, digits = 0); labelsD
+  
+  get_legend(ggplot(plotdf, aes_string(x = sprintf("factor(%s)", xvar_colname)
+                            , y = yvar_colname)) +
+    
+    geom_tile(aes(fill = .data[[lig_dist_colname]])
+              , colour = "white") +
+    scale_fill_gradient2(midpoint = lig_mean
+                         , low  = "green"
+                         , mid  = "yellow"
+                         , high = "red"
+                         , breaks = labels
+                         , limits = c(lig_min, lig_max)
+                         , labels = labelsD
+                         , name   = legend_title)
+    )
+}

scripts/functions/lineage_dist.R

@@ -16,7 +16,7 @@ lineage_distP <- function(plotdf
                           , all_lineages = F
                           , use_lineages = c("L1", "L2", "L3", "L4")
                           , with_facet = F
-                          , facet_wrap_var = ""
+                          , facet_wrap_var = "" # FIXME: document what this is for
                           , fill_categ = "mutation_info_labels"
                           , fill_categ_cols = c("#E69F00", "#999999")
                           , my_ats = 15 # axis text size
@@ -35,7 +35,7 @@ if(!all_lineages){
     plotdf = plotdf[plotdf[[y_axis]]%in%use_lineages,]
   }
   
-LinDistP = ggplot(plotdf, aes_string(x = x_axis
+  ggplot(plotdf, aes_string(x = x_axis
                             , y = y_axis))+
     
     geom_density_ridges(aes_string(fill = fill_categ)
@@ -57,19 +57,17 @@ LinDistP = ggplot(plotdf, aes_string(x = x_axis
           , legend.title = element_text(size = my_leg_title)
           , legend.position = c(0.8, 0.9)) +
     labs(x = x_lab
-       , fill = leg_label)
+         , fill = leg_label) + 
   
+  # FIXME: This didn't work BEFORE i fixed the ggplot() assignment thing!!!
   if (with_facet){
     
     # used reformulate or make as formula
     #fwv = reformulate(facet_wrap_var)
     fwv = as.formula(paste0("~", facet_wrap_var))
     
-  LinDistP = LinDistP + 
     facet_wrap(fwv) + 
       theme(legend.position = leg_pos_wf
             , legend.direction = leg_dir_wf)
   }
-
-return(LinDistP)
 }

scripts/functions/logoP_or.R

@@ -59,12 +59,16 @@ LogoPlotCustomH <- function(plot_df
                             , leg_dir = "horizontal" #can be vertical or horizontal
                             , leg_ts = 15 # leg text size
                             , leg_tts = 16 # leg title size
+                            , tpos0 = 0 # 0 is a magic number that does my sensible default
+                            , tW0 = 1
+                            , tH0 = 0.3
 )
 
 {
   #################################
   # Data processing for logo plot
   #################################
+  plot_df = generate_distance_colour_map(plot_df, yvar_colname = y_axis_colname, debug=TRUE)
   
   if (rm_empty_y){
     plot_df = plot_df[!is.na(plot_df[y_axis_colname]),]
@@ -142,7 +146,16 @@ LogoPlotCustomH <- function(plot_df
     ytt_col    = "black"
   }
   
-  p0 = ggseqlogo(logo_dfP_wf
+  if (y_axis_log){
+    
+    if (grepl("Log", y_lab)){
+      y_lab = y_lab
+      
+    }else{
+      y_lab = paste("Log", y_lab)
+    }
+  }
+  ggseqlogo(logo_dfP_wf
             , method = "custom"
             , col_scheme = my_logo_col
             , seq_type = "aa") + 
@@ -172,26 +185,20 @@ LogoPlotCustomH <- function(plot_df
     scale_x_discrete(x_lab
                      #, breaks
                      , labels = position_or
-                     , limits = factor(1:length(position_or)))
+                     , limits = factor(1:length(position_or))) +
     
-    LogoPlot = p0 + scale_y_continuous(y_lab
+    scale_y_continuous(y_lab
                        , breaks = seq(0, (y_lim), by = y_axis_increment)
                        #, labels = seq(0, (y_lim), by = y_axis_increment)
-                       , limits = c(0, y_lim))
+                       , limits = c(0, y_lim)) +
-    if (y_axis_log){
+    ylab(y_lab) +
-      
+    geom_tile(aes(plot_df$position, tpos0 # heat-mapped distance tiles along the bot
-      if (grepl("Log", y_lab)){
+                  , width  = tW0
-        y_lab = y_lab
+                  , height = tH0)
-        
+              , fill     = plot_df$ligD_colours
-      }else{
+              , colour   = plot_df$ligD_colours
-        y_lab = paste("Log", y_lab)
+              , linetype = "blank")
-      }
+  #LogoPlot = p0 + ylab(y_lab)
-      
+  #return(LogoPlot)
-    LogoPlot = p0 + ylab(y_lab)
-  
-    }
-
-
-  return(LogoPlot)
   
 }

scripts/functions/logoP_snp.R

@@ -48,6 +48,11 @@ LogoPlotSnps <- function(plot_df
                          , leg_dir = "horizontal" #can be vertical or horizontal
                          , leg_ts = 20 # leg text size
                          , leg_tts = 16 # leg title size
+                         , tpos0 = 0 # 0 is a magic number that does my sensible default
+                         , tW0 = 1
+                         , tH0 = 0.2
+                         , debug=FALSE
+                         
 )
 
 {
@@ -58,11 +63,15 @@ LogoPlotSnps <- function(plot_df
   ############################################
   # Data processing for logo plot for nsSNPS
   ############################################
-  setDT(plot_df)[, mut_pos_occurrence := .N, by = .(eval(parse(text=x_axis_colname)))] 
   
+  # Generate "ligand distance" colour map
+  plot_df = generate_distance_colour_map(plot_df, debug=TRUE)
+  
+  setDT(plot_df)[, mut_pos_occurrence := .N, by = .(eval(parse(text=x_axis_colname)))] 
+  if (debug) {
   table(plot_df[[x_axis_colname]])
   table(plot_df$mut_pos_occurrence)
-  
+  }
   max_mut = max(table(plot_df[[x_axis_colname]]))
   
   # Subset Data as specified by user
@@ -73,13 +82,14 @@ LogoPlotSnps <- function(plot_df
     my_data_snp = plot_df
     u = unique(my_data_snp[[x_axis_colname]])
     max_mult_mut = max(table(my_data_snp[[x_axis_colname]]))
-    
+    if (debug) {
       cat("\nNo filtering requested:"
           , "\nTotal no. of nsSNPs:", sum(table(plot_df$mut_pos_occurrence))
           , "\nTotal no. of nsSNPs omitted:", sum(table(plot_df$mut_pos_occurrence)[omit_snp_count])
           , "\nDim of data:", dim(my_data_snp)
           , "\nNo. of positions:", length(u)
           , "\nMax no. of muts at any position:", max_mult_mut)
+    }
   } else {
     
     my_data_snp = subset(plot_df, !(mut_pos_occurrence%in%omit_snp_count) )
@@ -88,7 +98,7 @@ LogoPlotSnps <- function(plot_df
     got_rows = sum(table(my_data_snp$mut_pos_occurrence))
     u = unique(my_data_snp[[x_axis_colname]])
     max_mult_mut = max(table(my_data_snp[[x_axis_colname]]))
-    
+    if (debug) {
       if (got_rows == exp_nrows) {
         cat("\nPass: Position with the stated nsSNP frequency filtered:", omit_snp_count
             , "\nTotal no. of nsSNPs:", sum(table(plot_df$mut_pos_occurrence))
@@ -103,6 +113,7 @@ LogoPlotSnps <- function(plot_df
             , "\nGot:", got_rows )
       }
     }
+  }
   
   #--------------------------------------
   # matrix for mutant type
@@ -116,19 +127,23 @@ LogoPlotSnps <- function(plot_df
   tab_mt = unclass(tab_mt)
   
   if (is.matrix(tab_mt)){
+    if (debug) {
       cat("\nPASS: Mutant matrix successfully created..."
           #, "\nRownames of mutant matrix:", rownames(tab_mt)
           #, "\nColnames of mutant matrix:", colnames(tab_mt)
       )
+    }
   } else{
     tab_mt = as.matrix(tab_mt, rownames = T)
     if (is.matrix(tab_mt)){
+      if (debug) {
         cat("\nCreating mutant matrix..."
             #, "\nRownames of mutant matrix:", rownames(tab_mt)
             #, "\nColnames of mutant matrix:", colnames(tab_mt)
         )
       }
     }
+  }
   
   #-------------------------------------
   # matrix for wild type
@@ -146,7 +161,7 @@ LogoPlotSnps <- function(plot_df
   wt
   
   tab_wt = table(wt[[symbol_wt_colname]], wt[[x_axis_colname]]); tab_wt # should all be 1
-  
+  if (debug) {
     if ( identical(colnames(tab_mt), colnames(tab_wt) ) && identical(ncol(tab_mt), ncol(tab_wt)) ){
       
       cat("\nPASS: Wild type matrix successfully created"
@@ -157,6 +172,7 @@ LogoPlotSnps <- function(plot_df
           #, "\nColnames of mutant matrix:", colnames(tab_wt)
       )
     }
+  }
   
   ######################################
   # Generating plots for muts and wt
@@ -188,11 +204,11 @@ LogoPlotSnps <- function(plot_df
   #####################################
   # Generating logo plots for nsSNPs
   #####################################
-  
+  cowplot::plot_grid(
     #-------------------
     # Mutant logo plot
     #-------------------
-  p0 = ggseqlogo(tab_mt
+    ggseqlogo(tab_mt
               , method = 'custom'
               , col_scheme = my_logo_col
               , seq_type = 'aa') + 
@@ -204,18 +220,21 @@ LogoPlotSnps <- function(plot_df
       scale_x_continuous(breaks = 1:ncol(tab_mt)
                          , expand = c(0.01,0)
                          , labels = colnames(tab_mt))+
+      
       scale_y_continuous(breaks = 0:(max_mult_mut-1)
                          , labels = c(1:max_mult_mut)
                          , limits = c(0, max_mult_mut)) +
-    #xlab(x_lab) +
+      # FIXME: currently broken, possibly due to ggseqlogo() not working in the
-    ylab(y_lab)
+      # way standard ggplot2() does
-
+      geom_tile(aes(plot_df$position, tpos0 # heat-mapped distance tiles along the bottom. 
-  cat('\nDone: p0')
+                    , width  = tW0
-  
+                    , height = tH0)
-  # further customisation
+                , fill     = plot_df$ligD_colours
-  mut_logo_p = p0 + theme(legend.position = leg_pos
+                , colour   = plot_df$ligD_colours
+                , linetype = "blank") +
+      ylab(y_lab) +
+      theme(legend.position = leg_pos
             , legend.direction = leg_dir
-                          #, legend.title = element_blank()
             , legend.title = element_text(size = leg_tts
                                           , colour = ytt_col)
             , legend.text = element_text(size = leg_ts)
@@ -225,7 +244,6 @@ LogoPlotSnps <- function(plot_df
                                          , hjust = 1
                                          , vjust = 0.4
                                          , colour = xfont_bgc)
-                          #, axis.text.y = element_blank()
             , axis.text.y = element_text(size = y_ats
                                          , angle = y_tangle
                                          , hjust = 1
@@ -236,16 +254,9 @@ LogoPlotSnps <- function(plot_df
             , axis.title.y = element_text(size = y_tts
                                           , colour = ytt_col)
             
-                          , plot.background = element_rect(fill = theme_bgc))
+            , plot.background = element_rect(fill = theme_bgc)
-  
+      )
-  cat('\nDone: mut_logo_p')
+    ,ggseqlogo(tab_wt
-  #return(mut_logo_p)
-  LogoPlotL[['mut_logoP']] <- mut_logo_p
-
-  #------------------
-  # Wild logo plot
-  #------------------
-  p1 = ggseqlogo(tab_wt
                , method = 'custom'
                , col_scheme = my_logo_col
                , seq_type = 'aa') + 
@@ -259,12 +270,7 @@ LogoPlotSnps <- function(plot_df
                          , expand = c(0.01,0)
                          , labels = colnames(tab_wt))+
       
-    xlab(x_lab)
+      xlab(x_lab) + 
-  
-  cat('\nDone: p1')
-  
-  # further customisation
-  wt_logo_p = p1 + 
       
       theme(legend.position = "none"
             , legend.direction = leg_dir
@@ -272,44 +278,22 @@ LogoPlotSnps <- function(plot_df
             , legend.title = element_text(size = y_tts
                                           , colour = ytt_col)
             , legend.text = element_text(size = leg_ts)
-          
             , axis.text.x = element_text(size = x_ats
                                          , angle = x_tangle
                                          , hjust = 1
                                          , vjust = 0.4
                                          , colour = xfont_bgc)
             , axis.text.y = element_blank()
-          
             , axis.title.x = element_text(size = x_tts
                                           , colour = xtt_col)
             , axis.title.y = element_text(size = y_tts
                                           , colour = ytt_col)
-          
+            , plot.background = element_rect(fill = theme_bgc)
-          , plot.background = element_rect(fill = theme_bgc))
+      )
-  
-  cat('\nDone: wt_logo_p')
-  #return(wt_logo_p)
-  LogoPlotL[['wt_logoP']] <- wt_logo_p
-  
-  #=========================================
-  # Output
-  # Combined plot: logo_mutliple_muts.svg
-  #=========================================
-  #suppressMessages( require(cowplot) )
-  #plot_grid(p1, p3, ncol = 1, align = 'v')
-  cat('\nDone: mut_logo_p + wt_logo_p')
-
-  # colour scheme: https://rdrr.io/cran/ggseqlogo/src/R/col_schemes.r
-  #cat("\nOutput plot:", LogoSNPs_comb, "\n")
-  #svg(LogoSNPs_combined, width = 32, height = 10)
-  LogoPlotL[['wt_logoP']] <- wt_logo_p
-  
-  LogoSNPs_comb = cowplot::plot_grid(LogoPlotL[['mut_logoP']]
-                                     , LogoPlotL[['wt_logoP']]
     , nrow = 2
     , align = "v"
     , rel_heights = c(3/4, 1/4))
-
+  #------------------
-  return(LogoSNPs_comb)
+  # Wild logo plot
-
+  #------------------
 }

scripts/functions/position_count_bp.R

@@ -105,8 +105,8 @@ site_snp_count_bp <- function (plotdf
   # not sure if to use with sort or directly
   my_x = sort(unique(snpsBYpos_df$snpsBYpos)) 
   
-  g = ggplot(snpsBYpos_df, aes(x = snpsBYpos))
+  ggplot(snpsBYpos_df, aes(x = snpsBYpos)) +
-  OutPlot_pos_count = g + geom_bar(aes (alpha = 0.5)
+    geom_bar(aes (alpha = 0.5)
              , show.legend = FALSE) +
     scale_x_continuous(breaks = unique(snpsBYpos_df$snpsBYpos)) +
     geom_label(stat = "count", aes(label = ..count..)
@@ -127,13 +127,10 @@ site_snp_count_bp <- function (plotdf
           , plot.subtitle = element_text(size = subtitle_size
                                          , hjust = 0.5
                                          , colour = subtitle_colour)) +
-    
     labs(title = bp_plot_title
          , subtitle = subtitle_text
          , x = xaxis_title
          , y = yaxis_title)
-
-  return(OutPlot_pos_count)
 }
 
 ########################################################################

scripts/functions/wideP_consurf.R

@@ -0,0 +1,506 @@
+#!/usr/bin/env Rscript 
+
+#########################################################
+# TASK: function for wide plot
+#with consurf score and error bars
+#position numbers coloured by 
+#    - ligand distance
+#    - active site residues
+#########################################################
+
+#==========================================================
+# wideP(): 
+# input args
+#==========================================================
+wideP_consurf <- function(plot_df
+                          , xvar_colname = "position"
+                          , yvar_colname = "consurf_score"
+                          , yvar_colourN_colname = "consurf_colour_rev" # num from 0-1
+                          , plot_error_bars = T
+                          , upper_EB_colname = "consurf_ci_upper"
+                          , lower_EB_colname = "consurf_ci_lower"
+                          
+                          , plot_type = "point" # default is point
+                          , point_colours
+                          , p_size  = 2 
+                          , leg_title1   = ""
+                          , leg_labels   = c("0": "Insufficient Data"
+                                             , "1": "Variable"
+                                             , "2", "3", "4", "5", "6", "7", "8"
+                                             , "9": "Conserved")
+                          , panel_col      = "black"
+                          , panel_col_fill = "black"
+                          
+                          # axes title and label sizes
+                          , x_axls = 12 # x-axis label size
+                          , y_axls = 15 # y-axis label size
+                          , x_axts = 12 # x-axis text size
+                          , y_axts = 12 # y-axis text size
+                          , default_xtc = "black" # x-axis text colour
+                          , ptitle = ""
+                          , xlab   = ""
+                          , ylab   = ""
+                          , pts    = 20
+                          
+                          # plot margins
+                          , t_margin    =  0.5
+                          , r_margin    =  0.5
+                          , b_margin    =  1
+                          , l_margin    =  1
+                          , unit_margin = "cm"
+                          
+                          # Custom 1: x-axis: text colour
+                          , xtext_colour_aa = F
+                          , xtext_colour_aa1 = active_aa_pos
+                          , xtext_colour_aa2 = aa_pos_drug
+                          , xtext_colours = c("purple", "brown", "black")
+                          
+                          # Custom 2: x-axis: geom tiles ~ lig distance
+                          , A_xvar_lig = T
+                          , leg_title2 = "Ligand Distance"
+                          , lig_dist_colname = LigDist_colname # from globals
+                          , lig_dist_colours = c("green", "yellow", "orange", "red")
+                          , tpos0 = 0 # 0 is a magic number that does my sensible default
+                          , tW0 = 1
+                          , tH0 = 0.3
+                          
+                          # Custom 3: x-axis: geom tiles ~ active sites and ligand
+                          , A_xvar_aa = F
+                          , aa_pos_drug = NULL
+                          , drug_aa_colour = "purple"
+                          , tW = 1
+                          , tH = 0.2
+                          , active_aa_pos = NULL
+                          , active_aa_colour = "brown"
+                          
+                          , aa_pos_lig1 = NULL
+                          , aa_colour_lig1 = "blue"
+                          , tpos1 = 0
+                          
+                          , aa_pos_lig2 = NULL
+                          , aa_colour_lig2 =  "cyan"
+                          , tpos2 = 0
+                          
+                          , aa_pos_lig3 = NULL
+                          , aa_colour_lig3 =  "cornflowerblue"
+                          , tpos3 = 0
+                          
+                          , default_gt_clr = "white"
+                          , debug=FALSE
+){
+  
+  if(missing(point_colours)){
+    temp_cols = colorRampPalette(c("seagreen", "sienna3"))(30)
+    point_colours = temp_cols
+  }else{
+    point_colours = point_colours
+  }
+  
+  ###############################
+  # custom 1: x-axis text colour
+  ##############################
+  
+  if (xtext_colour_aa){
+    positionF <- levels(as.factor(plot_df[[xvar_colname]]))
+    length(positionF)
+    aa_pos_colours = ifelse(positionF%in%xtext_colour_aa1, xtext_colours[1]
+                            , ifelse(positionF%in%xtext_colour_aa2
+                                     , xtext_colours[2]
+                                     , xtext_colours[3]))
+  }else{
+    aa_pos_colours = default_xtc 
+  }
+  
+  ################################################
+  # Custom 2: x-axis geom tiles ~ lig distance
+  ################################################
+  
+  #=========================
+  # Build data with colours
+  # ~ ligand distance
+  #=========================
+  if (A_xvar_lig){
+    cat("\nAnnotating x-axis ~", lig_dist_colname, "requested...")
+    
+    #-------------------------------------
+    # round column values: to colour by
+    #--------------------------------------
+    #plot_df = plot_df[order(plot_df[[lig_dist_colname]]),]
+    plot_df['lig_distR'] = round(plot_df[[lig_dist_colname]], digits = 0)
+    head(plot_df['lig_distR'])
+    
+    #-------------------------------------
+    # ligand distance range, min, max, etc
+    #--------------------------------------
+    lig_min  = min(round(plot_df[[lig_dist_colname]]), na.rm = T); lig_min
+    lig_max  = max(round(plot_df[[lig_dist_colname]]), na.rm = T); lig_max
+    lig_mean = round(mean(round(plot_df[[lig_dist_colname]]), na.rm = T)); lig_mean
+    
+    #-------------------------------------
+    # Create mapping colour key
+    #--------------------------------------
+    # sorting removes NA, so that n_colours  ==  length(ligD_valsR)
+    n_colours  = length(sort(unique(round(plot_df[[lig_dist_colname]], digits = 0)))); n_colours
+    
+    lig_cols = colorRampPalette(lig_dist_colours)(n_colours); lig_cols
+    ligD_valsR = sort(unique(round(plot_df[[lig_dist_colname]], digits = 0))); ligD_valsR
+    length(ligD_valsR)
+    
+    if (n_colours == length(ligD_valsR)) {
+      cat("\nStarting: mapping b/w"
+          , lig_dist_colname
+          , "and colours")
+    }else{
+      cat("\nCannot start mapping b/w", lig_dist_colname, "and colours..."
+          , "\nLength mismatch:"
+          , "No. of colours: ", n_colours
+          , "\nValues to map:", length(ligD_valsR))
+    }
+    
+    ligDcolKey <- data.frame(ligD_colours = lig_cols
+                             , lig_distR = ligD_valsR); ligDcolKey
+    names(ligDcolKey)
+    cat("\nSuccessful: Mapping b/w", lig_dist_colname, "and colours")
+    
+    #-------------------------------------
+    # merge colour key with plot_df
+    #--------------------------------------
+    plot_df = merge(plot_df, ligDcolKey, by = 'lig_distR')
+    
+    plot_df_check = as.data.frame(cbind(position    = plot_df[[xvar_colname]]
+                                       , ligD      = plot_df[[lig_dist_colname]]
+                                       , ligDR     = plot_df$lig_distR
+                                       , ligD_cols = plot_df$ligD_colours))
+  } else{
+    plot_df = plot_df
+  }
+  
+  ###############################################
+  # Custom 3: x-axis geom tiles ~ active sites
+  ################################################
+  
+  #==========================
+  # Build Data with colours
+  # ~ on active sites
+  #==========================
+  
+  if(A_xvar_aa) {
+    cat("\nAnnotation for xvar requested. Building colours for annotation...")
+    
+    aa_colour_colname    =  "bg_all"
+    aa_colour_colname1   =  "col_bg1"
+    aa_colour_colname2   =  "col_bg2"
+    aa_colour_colname3   =  "col_bg3"
+    
+    #--------------------------------------------------
+    # column colour 0: Active site + drug binding sites
+    #--------------------------------------------------
+    plot_df[[aa_colour_colname]] = ifelse(plot_df[[xvar_colname]]%in%aa_pos_drug
+                                         , drug_aa_colour
+                                         , ifelse(plot_df[[xvar_colname]]%in%active_aa_pos
+                                                  , active_aa_colour, default_gt_clr ))
+    plot_df[[aa_colour_colname]] 
+    cat("\nColumn created 'bg_all':", length(plot_df[[aa_colour_colname]]))
+    
+    #------------------------------------------------
+    # column colour 1: Ligand 1 + drug binding sites
+    #------------------------------------------------
+    cat("\nAssigning colours to drug binding and ligand-1 binding residues")
+    plot_df[[aa_colour_colname1]] = plot_df[[aa_colour_colname]]
+    plot_df[[aa_colour_colname1]] =  ifelse(plot_df[[xvar_colname]]%in%aa_pos_lig1
+                                           , aa_colour_lig1, plot_df[[aa_colour_colname]])
+    #------------------------------------------------
+    # column colour 2: Ligand 2
+    #------------------------------------------------
+    plot_df[[aa_colour_colname2]] = plot_df[[aa_colour_colname1]] 
+    plot_df[[aa_colour_colname2]] =  ifelse(plot_df[[xvar_colname]]%in%aa_pos_lig2
+                                           , aa_colour_lig2, plot_df[[aa_colour_colname1]])
+    
+    #------------------------------------------------
+    # column colour 3: Ligand 3
+    #------------------------------------------------
+    plot_df[[aa_colour_colname3]] = plot_df[[aa_colour_colname2]] 
+    plot_df[[aa_colour_colname3]] =  ifelse(plot_df[[xvar_colname]]%in%aa_pos_lig3
+                                           , aa_colour_lig3, plot_df[[aa_colour_colname2]])
+    
+  }  
+  ###################
+  # start plot
+  ###################
+  
+  #-------------------
+  # x and y axis
+  # range, scale, etc
+  #-------------------
+  my_xlim = length(unique(plot_df[[xvar_colname]])); my_xlim
+  ymin    = min(plot_df[[yvar_colname]]); ymin
+  ymax    = max(plot_df[[yvar_colname]]); ymax
+  
+  g = ggplot(plot_df, aes_string(x = sprintf("factor(%s)", xvar_colname)
+                                , y = yvar_colname
+                                , colour = sprintf("factor(%s)", yvar_colourN_colname)
+  ))
+  
+  "if SPECIAL do SPECIAL THING, otherwise do NORMAL THING"
+  if (plot_type == "bar"){
+    g0 = g +
+      geom_bar(stat = "identity")
+  }
+  else{
+    g0 = g +
+      coord_cartesian(xlim = c(1, my_xlim)
+                      , ylim = c(ymin, ymax)
+                      , clip = "off") +
+      geom_point(size = p_size) +
+      scale_colour_manual(values   = point_colours)
+  }
+  
+  if (plot_error_bars){
+    g0 = g0 +
+      geom_errorbar(aes(ymin = eval(parse(text = lower_EB_colname))
+                        , ymax = eval(parse(text = upper_EB_colname))
+      ))
+  }else{
+    
+    g0 = g0
+    
+  }
+  
+  #---------------------
+  # add axis formatting
+  #---------------------
+  g1 = g0 + theme( axis.text.x = element_text(size = x_axts
+                                              , angle = 90
+                                              , hjust = 1
+                                              , vjust = 0.4
+                                              , face = "bold"
+                                              , colour = aa_pos_colours)
+                   , axis.text.y = element_text(size = y_axts 
+                                                , angle = 0
+                                                , hjust = 1
+                                                , vjust = 0)
+                   , axis.title.x = element_text(size = x_axls)
+                   , axis.title.y = element_text(size = y_axls )
+                   , panel.background = element_rect(fill = panel_col_fill, color = panel_col)
+                   , panel.grid.major = element_line(color = "black")
+                   , panel.grid.minor = element_line(color = "black")
+                   , plot.title       = element_text(size = pts
+                                                     , hjust = 0.5)
+                   , plot.margin = margin(t = t_margin
+                                          , r = r_margin
+                                          , b = b_margin
+                                          , l = l_margin
+                                          , unit = unit_margin))+
+    guides(colour = guide_legend(title = "ConsurfXXXX")) +
+    
+    labs(title = ptitle
+         , x = xlab
+         , y = ylab)
+  
+  #------------------
+  #Extract legend1
+  #------------------
+  # yayy
+  g1_leg = ggplot(plot_df, aes_string(x = sprintf("factor(%s)"
+                                                 , xvar_colname) ))
+  g1_leg = g1_leg + geom_bar(); g1_leg
+  g1_leg = g1_leg + geom_bar(aes_string(fill = sprintf("factor(%s)"
+                                                       , yvar_colourN_colname))) 
+  
+  g1_leg = g1_leg + scale_fill_manual(values = consurf_palette2 , name = leg_title1) 
+  g1_leg
+  
+  legend1 = get_legend(g1_leg)
+  
+  
+  #####################################################
+  #============================================
+  # x-axis: geom_tiles ~ ligand distance
+  #============================================
+  #-------
+  # plot
+  #-------
+  if(A_xvar_lig){ # 0 is a magic number that does my sensible default
+    if (tpos0 == 0){
+      tpos0 = ymin-0.5
+    }
+    if (tpos1 == 0){
+      tpos1 = ymin-0.65
+    }
+    if (tpos2 == 0){
+      tpos2 = ymin-0.75
+    }
+    if (tpos3 == 0){
+      tpos3 = ymin-0.85
+    }
+    
+    
+    cat("\nColouring x-axis aa based on", lig_dist_colname
+        , "\nNo. of colours:", n_colours)
+    
+    g2 = g1 + geom_tile(aes(, 2
+#    g2 = g1 + geom_tile(aes(, tpos0
+
+                            , width  = tW0
+                            , height = tH0)
+                        , fill     = plot_df$ligD_colours
+                        , colour   = plot_df$ligD_colours
+                        , linetype = "blank")
+    
+    #cat("Nrows of plot df", length(plot_df$ligD_colours))
+    out = g2
+
+  }else{
+    out = g1
+  }    
+  #==============================================
+  # x-axis: geom_tiles ~ active sites and others
+  #==============================================
+  if(A_xvar_aa){
+    #tpos = 0
+    #tW  = 1
+    #tH  = 0.2 
+    
+    #---------------------
+    # Add2plot: 3 ligands
+    #---------------------
+    if (all(!is.null(active_aa_pos) &&
+            !is.null(aa_pos_drug) &&
+            !is.null(aa_pos_lig1) && !is.null(aa_pos_lig2) && !is.null(aa_pos_lig3))) {
+      if (debug){
+        cat("\n\nAnnotating xvar with active, drug binding, and Lig 1&2&3 sites")
+        cat("\nCreating column colours, column name:", aa_colour_colname3)
+        
+        cat("\nDoing Plot with 3 ligands")
+      }
+      out = out +  geom_tile(aes(,tpos3
+                                 , width  = tW
+                                 , height = tH )
+                             , fill = plot_df[[aa_colour_colname3]]
+                             , colour = plot_df[[aa_colour_colname3]]
+                             , linetype = "solid") +
+        geom_tile(aes(, tpos2
+                      , width  = tW
+                      , height = tH )
+                  , fill = plot_df[[aa_colour_colname2]]
+                  , colour = plot_df[[aa_colour_colname2]]
+                  , linetype = "solid")+
+        
+        geom_tile(aes(, tpos1
+                      , width  = tW
+                      , height = tH)
+                  , fill = plot_df[[aa_colour_colname1]]
+                  , colour = plot_df[[aa_colour_colname1]]
+                  , linetype = "solid")
+      if (debug){
+        cat("\nDone Plot with 3 ligands")
+      }
+    }
+    #---------------------
+    # Add2plot: 2 ligands
+    #---------------------
+    
+    if (all(!is.null(active_aa_pos) &&
+            !is.null(aa_pos_drug) &&
+            !is.null(aa_pos_lig1) && !is.null(aa_pos_lig2) && is.null(aa_pos_lig3))) {
+      if (debug){
+        cat("\n\nAnnotating xvar with active, drug binding, and Lig 1&2 sites")
+        cat("\nCreating column colours, column name:", aa_colour_colname2)
+        
+        cat("\nDoing Plot with 2 ligands")
+      }
+      out = out + 
+        geom_tile(aes(, tpos2
+                      , width  = tW
+                      , height = tH)
+                  , fill = plot_df[[aa_colour_colname2]]
+                  , colour = plot_df[[aa_colour_colname2]]
+                  , linetype = "solid")+
+        geom_tile(aes(, tpos1
+                      , width  = tW
+                      , height = tH)
+                  , fill = plot_df[[aa_colour_colname1]]
+                  , colour = plot_df[[aa_colour_colname1]]
+                  , linetype = "solid")
+      if (debug){
+        cat("\nDone Plot with 2 ligands")
+      }
+    }
+    
+    #---------------------
+    # Add2plot: 1 ligand
+    #---------------------
+    if (all(!is.null(active_aa_pos) &&
+            !is.null(aa_pos_drug) &&
+            !is.null(aa_pos_lig1) && is.null(aa_pos_lig2) && is.null(aa_pos_lig3))) {
+      if (debug){
+        cat("\n\nAnnotating xvar with active, drug binding, and Lig 1 sites")
+        cat("\nCreating column colours, column name:", aa_colour_colname1)
+        
+        cat("\nDoing Plot with 1 ligands")
+      }
+      out = out + 
+        geom_tile(aes(, tpos1
+                      , width  = tW
+                      , height = tH)
+                  , fill = plot_df[[aa_colour_colname1]]
+                  , colour = plot_df[[aa_colour_colname1]]
+                  , linetype = "solid")
+      
+      cat("\nDone Plot with 1 ligand")
+      
+    }
+    
+    #-----------------------------------
+    # Add2plot:NO ligands
+    # No Ligs: Just drug and active site
+    # DEFAULT: A_xvar_aa == TRUE
+    #----------------------------------
+    if (all(!is.null(active_aa_pos) &&
+            !is.null(aa_pos_drug) &&
+            is.null(aa_pos_lig1) &&
+            is.null(aa_pos_lig2) &&
+            is.null(aa_pos_lig3))) {
+      if (debug){
+        cat("\n\nAnnotating xvar with active and drug binding sites")
+        cat("\nCreating column colours, column name:", aa_colour_colname)
+        cat("\nDoing Plot with 0 ligands: active and drug site only")
+      }
+      out = out + geom_tile(aes(, tpos3
+                                , width  = tW
+                                , height = tH)
+                            , fill = plot_df[[aa_colour_colname]]
+                            , colour = plot_df[[aa_colour_colname]]
+                            , linetype = "solid")
+      if (debug){
+        cat("\nDone Plot with: Active and Drug sites")
+      }
+    }
+  }else{
+    cat("\nNo annotation for additional ligands on xvar requested")
+  }
+  if (A_xvar_lig){
+    legs = cowplot::plot_grid(legend1
+                              , generate_distance_legend(plot_df, yvar_colname = yvar_colname)
+                              , ncol = 1
+                              , align = "hv"
+                              , rel_heights = c(2/4,3/4)) 
+    
+    out2 = cowplot::plot_grid( out + theme(legend.position = "none")
+                               , legs
+                               , ncol = 2
+                               , align = "hv"
+                               , rel_widths = c(9/10, 0.4/10)
+    )
+  }else{
+    out2 = cowplot::plot_grid( out + theme(legend.position = "none")
+                               , legend1
+                               , ncol = 2
+                               , align = "hv"
+                               , rel_widths = c(9/10, 0.5/10)
+    )
+  }
+  return(out2)
+  #return(out2) 
+  
+}