fix #1043 and address #1053

PoisonAlien · Sep 27, 2024 · 95e5e04 · 95e5e04
1 parent de9e59f
commit 95e5e04
Show file tree

Hide file tree

Showing 6 changed files with 124 additions and 124 deletions.
diff --git a/R/gisticOncoPlot.R b/R/gisticOncoPlot.R
@@ -39,7 +39,7 @@
 
 
 gisticOncoPlot = function(gistic = NULL, top = NULL, bands = NULL,
-                           showTumorSampleBarcodes = FALSE, gene_mar = 5, barcode_mar = 6, sepwd_genes = 0.5, sepwd_samples = 0.25, clinicalData = NULL, clinicalFeatures = NULL, sortByAnnotation = FALSE, sampleOrder = NULL,
+                           showTumorSampleBarcodes = FALSE, gene_mar = 5, barcode_mar = 6, right_mar = 2.5, sepwd_genes = 0.5, sepwd_samples = 0.25, clinicalData = NULL, clinicalFeatures = NULL, sortByAnnotation = FALSE, sampleOrder = NULL,
                            annotationColor = NULL, bandsToIgnore = NULL,
                            removeNonAltered = TRUE, colors = NULL, SampleNamefontSize = 0.6, fontSize = 0.8, legendFontSize = 1.2, annotationFontSize = 1.2, borderCol = "white", bgCol = "#CCCCCC") {
 
@@ -133,15 +133,15 @@ gisticOncoPlot = function(gistic = NULL, top = NULL, bands = NULL,
 
   if(is.null(clinicalFeatures)){
     if(showTumorSampleBarcodes){
-      par(mar = c(barcode_mar, gene_mar, 1, 2.5), xpd = TRUE)
+      par(mar = c(barcode_mar, gene_mar, 1, right_mar), xpd = TRUE)
     }else{
-      par(mar = c(0.5, gene_mar, 1, 2.5), xpd = TRUE)
+      par(mar = c(0.5, gene_mar, 1, right_mar), xpd = TRUE)
     }
   }else{
     if(showTumorSampleBarcodes){
-      par(mar = c(barcode_mar, gene_mar, 1, 5), xpd = TRUE)
+      par(mar = c(barcode_mar, gene_mar, 1, right_mar), xpd = TRUE)
     }else{
-      par(mar = c(0.5, gene_mar, 1, 5), xpd = TRUE)
+      par(mar = c(0.5, gene_mar, 1, right_mar), xpd = TRUE)
     }
   }
 
@@ -207,7 +207,7 @@ gisticOncoPlot = function(gistic = NULL, top = NULL, bands = NULL,
 
     annotation = annotation[colnames(numMat), ncol(annotation):1, drop = FALSE]
 
-    par(mar = c(0, gene_mar, 0, 5), xpd = TRUE)
+    par(mar = c(0, gene_mar, 0, right_mar), xpd = TRUE)
 
     image(x = 1:nrow(annotation), y = 1:ncol(annotation), z = as.matrix(annotation),
           axes = FALSE, xaxt="n", yaxt="n", bty = "n",

diff --git a/R/oncomatrix.R b/R/oncomatrix.R
@@ -167,6 +167,7 @@ sortByAnnotation <-function(numMat,maf, anno, annoOrder = NULL, group = TRUE, is
     #anno.spl = split(anno, as.numeric(as.character(anno[,1]))) #sorting only first annotation
     mat_samps = colnames(numMat)[colnames(numMat) %in% rownames(anno)]
     anno = anno[mat_samps,, drop = FALSE]
+    anno[,1] = as.numeric(as.character(anno[,1]))
     anno = anno[order(anno[,1], na.last = TRUE),, drop = FALSE]
     numMat.sorted = numMat[,rownames(anno)]
   }else{
@@ -575,32 +576,40 @@ parse_annotation_dat = function(annotationDat = NULL, clinicalFeatures = NULL){
   return(annotation)
 }
 
-get_anno_cols = function(ann, numericAnnoCol = NULL){
+get_anno_cols = function(ann, numericAnnoCol = "YlOrBr"){
   ann_cols = list()
-  if(is.null(numericAnnoCol)){
-    numericAnnoCol =  RColorBrewer::brewer.pal(n = 9, name = "YlOrBr")
-  }else{
-    numericAnnoCol =  RColorBrewer::brewer.pal(n = 9, name = numericAnnoCol)
-  }
+  col_class = c()
+
   for(i in 1:ncol(ann)){
     if(is.numeric(ann[,i])){
-      x = ann[,i]
-      ann_lvls_cols = colorRampPalette(numericAnnoCol)(length(x))
-      names(ann_lvls_cols) = x[order(x, na.last = TRUE)]
+      x = as.numeric(as.character(ann[,i]))
+      x_unique = unique(x)
+      numericAnnoCol = sample(x = c("Blues", "BuGn", "BuPu", "GnBu", "Greens", "Greys", "Oranges",
+        "OrRd", "PuBu", "PuBuGn", "PuRd", "Purples", "RdPu", "Reds",
+        "YlGn", "YlGnBu", "YlOrBr", "YlOrRd"), size = 1)
+      #message("Using ", numericAnnoCol)
+      numericAnnoCol =  RColorBrewer::brewer.pal(n = 9, name = numericAnnoCol)
+      ann_lvls_cols = colorRampPalette(numericAnnoCol)(length(x_unique))
+      names(ann_lvls_cols) = x_unique[order(x_unique, na.last = TRUE)]
+      ann_lvls_cols = ann_lvls_cols[as.character(x)]
       ann_cols[[i]] = ann_lvls_cols
+      col_class = c(col_class, "numeric")
     }else{
       ann_lvls = unique(as.character(ann[,i]))
       if(length(ann_lvls) <= 9){
         ann_lvls_cols = RColorBrewer::brewer.pal(n = 9, name = 'Set1')[1:length(ann_lvls)]
       }else{
         ann_lvls_cols = colors()[sample(x = 1:100, size = length(ann_lvls), replace = FALSE)]
       }
+      names(ann_lvls_cols) = ann_lvls
+      ann_lvls_cols = ann_lvls_cols[as.character(ann[,i])]
       ann_cols[[i]] = ann_lvls_cols
-      names(ann_cols[[i]]) = ann_lvls
+      col_class = c(col_class, "char")
     }
   }
 
   names(ann_cols) = colnames(ann)
+  attr(ann_cols, "class") = col_class
 
   return(ann_cols)
 }

diff --git a/R/oncoplot.R b/R/oncoplot.R
@@ -962,55 +962,55 @@ oncoplot = oncoplot = function(maf, top = 20, minMut = NULL, genes = NULL, alter
   #06: Plot annotations if any
   if(!is.null(clinicalFeatures)){
 
-    #clini_lvls = as.character(unlist(lapply(annotation, function(x) unique(as.character(x)))))
-    clini_lvls = lapply(annotation, function(x) unique(as.character(x)))
+    annotation = annotation[colnames(numMat), ncol(annotation):1, drop = FALSE]
 
     if(is.null(annotationColor)){
+      #In case no-color codes are provided
       annotationColor = get_anno_cols(ann = annotation)
-    }
-    annotationColor = annotationColor[colnames(annotation)]
-
-    annotationColor = lapply(annotationColor, function(x) {
-      na_idx = which(is.na(names(x)))
-      x[na_idx] = "gray70"
-      names(x)[na_idx] = "NA"
-      x
-    })
-
-    anno_cols = c()
-    for(i in 1:length(annotationColor)){
-      anno_cols = c(anno_cols, annotationColor[[i]])
-    }
-
-    #clini_lvls = clini_lvls[!is.na(clini_lvls)]
-    clini_lvls = lapply(clini_lvls, function(cl){
-      temp_names = suppressWarnings(sample(
-        x = setdiff(x = 1:1000, y = as.numeric(as.character(cl))),
-        size = length(cl),
-        replace = FALSE
-      ))
-      names(cl) = temp_names#1:length(clini_lvls)
-      cl
-    })
+    }else{
+      #If provided - generate a continuous palette for continuous variables
+      annotationColor = annotationColor[colnames(annotation)]
+
+      annotationColor_temp = lapply(seq_along(annotationColor), function(idx){
+        if(is.numeric(annotation[,idx])){
+          #print(annotationColor[[idx]])
+          seq_col_pal = c("Blues", "BuGn", "BuPu", "GnBu", "Greens", "Greys", "Oranges",
+                          "OrRd", "PuBu", "PuBuGn", "PuRd", "Purples", "RdPu", "Reds",
+                          "YlGn", "YlGnBu", "YlOrBr", "YlOrRd")
+          col_match = annotationColor[idx] %in% seq_col_pal
+
+          if(!col_match){
+            stop("numeric annotation color for ", names(annotationColor)[idx] , " must be a sequential color palette!\n", paste(seq_col_pal, collapse = " "))
+          }
+          x = annotation[,idx]
+          x_unique = unique(x)
+          numericAnnoCol = RColorBrewer::brewer.pal(n = 9, name = annotationColor[[idx]])
+          ann_lvls_cols = colorRampPalette(numericAnnoCol)(length(x_unique))
+          names(ann_lvls_cols) = x_unique[order(x_unique, na.last = TRUE)]
+          ann_lvls_cols = ann_lvls_cols[as.character(x)]
+        }else{
+          ann_lvls_cols = annotationColor[[idx]]
+          ann_lvls_cols = ann_lvls_cols[as.character(annotation[,idx])]
+        }
+        ann_lvls_cols
+      })
+      names(annotationColor_temp) = names(annotationColor)
 
-    temp_rownames = rownames(annotation)
-    annotation = data.frame(lapply(annotation, as.character),
-                            stringsAsFactors = FALSE, row.names = temp_rownames)
-
-    annotation_color_coded = data.frame(row.names = rownames(annotation), stringsAsFactors = FALSE)
-    for(i in 1:length(clini_lvls)){
-      cl = clini_lvls[[i]]
-      clname = names(clini_lvls)[i]
-      cl_vals = annotation[,clname] #values in column
-      for(i in 1:length(cl)){
-        cl_vals[cl_vals == cl[i]] = names(cl[i])
+      #dirty way to check which columns are of numeric
+      temp_col_class = c()
+      for(i in 1:ncol(annotation)){
+        if(is.numeric(annotation[,i])){
+          temp_col_class = c(temp_col_class, "numeric")
+        }else{
+          temp_col_class = c(temp_col_class, "char")
+        }
       }
-      annotation_color_coded = cbind(annotation_color_coded, cl_vals, stringsAsFactors = FALSE)
+      annotationColor = annotationColor_temp
+      attr(annotationColor, "class") = temp_col_class
     }
-    names(annotation_color_coded) = names(clini_lvls)
 
-    annotation = data.frame(lapply(annotation_color_coded, as.numeric), stringsAsFactors=FALSE, row.names = temp_rownames)
-    annotation = annotation[colnames(numMat), ncol(annotation):1, drop = FALSE]
+    ann_classes = attributes(annotationColor)$class
+    #return(list(annotation, annotationColor))
 
     if(!is.null(leftBarData)){
       plot.new()
@@ -1022,40 +1022,43 @@ oncoplot = oncoplot = function(maf, top = 20, minMut = NULL, genes = NULL, alter
       par(mar = c(0, gene_mar, 0, 3), xpd = TRUE)
     }
 
-    image(x = 1:nrow(annotation), y = 1:ncol(annotation), z = as.matrix(annotation),
+    image(x = 1:nrow(annotation), y = 1:ncol(annotation), z = matrix(data = 0, nrow = nrow(annotation), ncol = ncol(annotation)),
           axes = FALSE, xaxt="n", yaxt="n", bty = "n",
-          xlab="", ylab="", col = "white") #col = "#FC8D62"
-
-    #Plot for all variant classifications
-    for(i in 1:length(clini_lvls)){
-      cl = clini_lvls[[i]]
-      clnames = names(clini_lvls)[i]
-      cl_cols = annotationColor[[clnames]]
-      for(i in 1:length(names(cl))){
-        anno_code = cl[i]
-        col = cl_cols[anno_code]
-        temp_anno = as.matrix(annotation)
-        #Handle NA's
-        if(is.na(col)){
-          col = "gray70"
-          temp_anno[is.na(temp_anno)] = as.numeric(names(anno_code))
-        }
-        temp_anno[temp_anno != names(anno_code)] = NA
+          xlab="", ylab="", col = "gray90") #col = "#FC8D62"
 
+    for(i in 1:ncol(annotation)){
+      temp_anno = annotation
+
+      if(ncol(temp_anno) > 1){
+        rm_idx = setdiff(y = i, x = 1:ncol(temp_anno))
+        temp_anno[,rm_idx] = NA
+      }
+
+      if(ann_classes[i] == "numeric"){
+        temp_anno = apply(temp_anno, 2, as.numeric)
         suppressWarnings(image(x = 1:nrow(temp_anno), y = 1:ncol(temp_anno), z = temp_anno,
-                               axes = FALSE, xaxt="n", yaxt="n", xlab="", ylab="", col = col, add = TRUE))
+                               axes = FALSE, xaxt="n", yaxt="n", xlab="", ylab="", col = annotationColor[[i]], add = TRUE))
+      }else{
+        temp_lvls = unique(names(annotationColor[[i]])) #unique(temp_anno[,i])
+        temp_lvls = temp_lvls[complete.cases(temp_lvls)] #Remove NAs
+        temp_lvls_cols = annotationColor[[i]][temp_lvls]
+        temp_anno[,i] = match(x = temp_anno[,i], table = unique(temp_lvls), nomatch = NA)
+        temp_anno = apply(temp_anno, 2, as.numeric)
+        suppressWarnings(image(x = 1:nrow(temp_anno), y = 1:ncol(temp_anno), z = temp_anno,
+                               axes = FALSE, xaxt="n", yaxt="n", xlab="", ylab="", col = temp_lvls_cols, add = TRUE))
       }
     }
 
-    #Add grids
-    rect(xleft = 0.5, ybottom = (0:(ncol(annotation))) + 0.5, xright = nrow(annotation)+0.5, ytop = (0:(ncol(annotation))) + 0.5, border = annoBorderCol, lwd = 0.7)
-    abline(v = (0:(nrow(nm))) + 0.5, col = annoBorderCol, lwd = sepwd_samples)
-    mtext(text = colnames(annotation), side = 4,
-          font = 1, line = 0.4, cex = fontSize, las = 2, at = 1:ncol(annotation))
+      #Add grids
+      rect(xleft = 0.5, ybottom = (0:(ncol(annotation))) + 0.5, xright = nrow(annotation)+0.5, ytop = (0:(ncol(annotation))) + 0.5, border = annoBorderCol, lwd = 0.7)
+      abline(v = (0:(nrow(nm))) + 0.5, col = annoBorderCol, lwd = sepwd_samples)
+      mtext(text = colnames(annotation), side = 4,
+            font = 1, line = 0.4, cex = fontSize, las = 2, at = 1:ncol(annotation))
+
+      if(drawRowBar){
+        plot.new()
+      }
 
-    if(drawRowBar){
-      plot.new()
-    }
   }
 
   #07: Draw TiTv plot
@@ -1161,32 +1164,29 @@ oncoplot = oncoplot = function(maf, top = 20, minMut = NULL, genes = NULL, alter
 
   x_axp = 0+lep$rect$w
 
-  #print(annotation)
-  #return(annotationColor)
-
   if(!is.null(clinicalFeatures)){
     #Check which of the clinicalFeatures are of numeric
-    ann_classes = lapply(names(annotationColor), function(ac){
-      temp_anno = data.table::copy(maf@clinical.data)
-      data.table::setDF(temp_anno)
-      ac_idx = which(colnames(temp_anno) == ac)
-      is.numeric(temp_anno[,ac_idx])
-    })
+    ann_classes = attributes(annotationColor)$class
     num_start = 0.1 #numeric legend starts from here
     num_width = 0.1
     num_incr = 0.2
 
-    for(i in 1:ncol(annotation)){
+    for(i in seq_along(ann_classes)){
       #x = unique(annotation[,i])
+      temp_lvls = unique(names(annotationColor[[i]]))
+      temp_lvls = temp_lvls[complete.cases(temp_lvls)]
+      temp_lvls_cols = annotationColor[[i]][temp_lvls]
       x = annotationColor[[i]]
-      is_num = ann_classes[[i]]
+      is_num = ann_classes[i] == "numeric"
       xt = names(x)
 
       if(is_num){
-        x_range = as.numeric(names(x))
-        x_range = x_range[!is.na(x_range)]
-        x_range = x_range[!is.infinite(x_range)]
-        x_range = range(x_range)
+        xt = as.numeric(xt)
+        xt = xt[!is.na(xt)]
+        xt = xt[!is.infinite(xt)]
+        xt = sort(xt)
+        x = x[as.character(xt)]
+        x_range = round(range(xt), 2)
 
         image(
           y = c(0.1, 0.2),
@@ -1195,31 +1195,21 @@ oncoplot = oncoplot = function(maf, top = 20, minMut = NULL, genes = NULL, alter
           col = x, xlim = c(0, 1), ylim = c(0, 1), axes = FALSE, xlab = NA, ylab = NA, add = TRUE
         )
         text(x = c(num_start, num_start+num_width), y = 0.2, labels = x_range, pos = 3)
-        text(x = num_start+(num_width/2), y = 0.1, labels = clinicalFeatures[i], pos = 1, adj = 1)
+        text(x = num_start+(num_width/2), y = 0.1, labels = names(annotationColor)[i], pos = 1, adj = 1)
         num_start = num_start + num_incr
 
       }else{
-        if("NA" %in% xt){
-          xt = xt[!xt %in% "NA"]
-          xt = sort(xt, decreasing = FALSE)
-          xt = c(xt, "NA")
-          x = x[xt]
-        }else{
-          xt = sort(xt, decreasing = FALSE)
-          x = x[xt]
-        }
 
-        if(length(x) <= 4){
+        if(length(temp_lvls_cols) <= 4){
           n_col = 1
         }else{
-          n_col = (length(x) %/% 4)+1
+          n_col = (length(temp_lvls_cols) %/% 4)+1
         }
-        names(x)[is.na(names(x))] = "NA"
 
-        lep = legend(x = x_axp, y = 1, legend = names(x),
-                     col = x, border = NA,
+        lep = legend(x = x_axp, y = 1, legend = names(temp_lvls_cols),
+                     col = temp_lvls_cols, border = NA,
                      ncol= n_col, pch = 15, xpd = TRUE, xjust = 0, bty = "n",
-                     cex = annotationFontSize, title = rev(names(annotation))[i],
+                     cex = annotationFontSize, title = names(annotationColor)[i],
                      title.adj = 0)
         x_axp = x_axp + lep$rect$w
 
@@ -1229,12 +1219,6 @@ oncoplot = oncoplot = function(maf, top = 20, minMut = NULL, genes = NULL, alter
 
   n_mut_samps = length(which(colSums(numMat) != 0))
   altStat = paste0("Altered in ", n_mut_samps, " (", round(n_mut_samps/totSamps, digits = 4)*100, "%) of ", totSamps, " samples.")
-  # if(removeNonMutated){
-  #   #mutSamples = length(unique(unlist(genesToBarcodes(maf = maf, genes = rownames(mat), justNames = TRUE))))
-  # }else{
-  #   mutSamples = length(unique(unlist(genesToBarcodes(maf = maf, genes = rownames(numMat), justNames = TRUE, verbose = FALSE))))
-  #   altStat = paste0("Altered in ", mutSamples, " (", round(mutSamples/totSamps, digits = 4)*100, "%) of ", totSamps, " samples.")
-  # }
 
   if(showTitle){
     if(is.null(titleText)){

diff --git a/R/readSegs.R b/R/readSegs.R
@@ -160,7 +160,7 @@ plotCBS = function(segData, tsb, build = 'hg19', chr.colors = NULL, y_lims = NUL
   chr.labels= c(1:22, 'X', 'Y')
   chr.lvls = levels(seg.spl.transformed[,Chromosome])
 
-  print(chr.colors)
+  #print(chr.colors)
   if(is.null(chr.colors)){
     chr.colors = c('gray70', 'midnightblue')
   }
@@ -171,6 +171,8 @@ plotCBS = function(segData, tsb, build = 'hg19', chr.colors = NULL, y_lims = NUL
   data.table::setkey(seg.spl.transformed, "Chromosome")
   N <- seg.spl.transformed[levels(Chromosome), .N, .EACHI][, N]
 
+  colnames(seg.spl.transformed)[1:6] = c("Sample", "Chromosome",  "Start_Position",  "End_Position", "Num_Probes", "Segment_Mean")
+
   if(is.null(y_lims)){
     y_lims = pretty(round(range(seg.spl.transformed$Segment_Mean, na.rm = TRUE, finite = TRUE), digits = 2))
   }

diff --git a/inst/NEWS.md b/inst/NEWS.md
@@ -9,14 +9,18 @@
 - MAJOR: `read.maf` by default coerces clinical data columns to character. This bug fix avoids it and is auto detected. Issue: [997](https://github.com/PoisonAlien/maftools/issues/997)
 
 ## ENHANCEMENTS
+- Better handling of color codes for continuous variable annotations [1053](https://github.com/PoisonAlien/maftools/issues/1053)
+- Add `right_mar` to `gisticOncoPlot`[1043](https://github.com/PoisonAlien/maftools/issues/1043)
 - Added `PPDPFL` to protein domain database Issue: [1025](https://github.com/PoisonAlien/maftools/issues/1025)
 - Better sorting of oncoplot with `collapsePathway`
 - Changed default background for oncoplot from `gray` to `#ecf0f1`
 - Changed default signature database to SBS_v3.4 (from legacy)
-- Update tmb function
+- Update `tmb` function
+
 
 ## NEW FUNCTIONS
 - `gisticCompare()` for comparing two GISTIC objects
+- `segSummarize()` for summarizing DNAcopy segments
 
 # CHANGES IN VERSION 2.18.0
 (Bioconductor release branch - 2.20.0)