CHANGES in VERSION 1.99.0
 
 This a major update of the package. The main changes are:
 * support column split
 * support align heatmaps vertically
 * add a naive `AnnotationFunction` class to handle annotation functions
 
 =======================
 

 CHANGES in VERSION 1.19.1
 
 * `Heatmap()`: no column name added if the input matrix is a one-column matrix.

 * `oncoPrint()`: scales the the row annotations are now the same if rows are split.

 
 =======================
 
 CHANGES in VERSION 1.17.1

 
 * `Legend()`: add `by_row` argument to control the arrangement of legends
   if they are put in more than one columns

 * `Legend()`: use `textGrob()` if the point symbol is text

 * `grid.dendrogram()`: fix a bug that the dendrogram is wrong when row/column
   names have duplicated names.

 * `anno_boxplot()`: axis rescaled when outline = FALSE

 * `oncoPrint()`: rows are first ordered by total number of mutations and then
   ordered by number of samples that have mutations

 * correctly reorder rows

 * add `row_gap` argument for list of heatmaps

 * `oncoPrint()`: add `j` and `i` as optional argument for `alter_fun`

 
 =======================
 

 CHANGES in VERSION 1.15.1
 
 * random colors are generated by new `rand_color()` function in circlize package.

 * add `density_param` in `densityHeatmap()` function
 * annotations with duplicated names have no legends any more
 * re-implement `grid.xaxis()` to draw axis labels rotated 90 degrees

 * grids in discrete legend are arranged by rows if ncol > 1

 * raster image is generated in an independent R session

 * empty string in annotation or heatmap is mapped to NA
 * annotation and heatmap legends can be merged into one column. 

 * change the default value of `row_names_max_width` and `column_names_max_height`

 * default legend style for continuous values is changed to "continuous"
 * add `grid.dendrogram2()` which draws dendrograms with uneven position for leaves
 * move **dendextend** to Suggests field because it depends/imports rlang indirectly
   which has a `print.frame()` function and it will affect to print a `frame` object
   returned by `frameGrob()`.
 * `decorate_*()` functions return to the viewport where they are called.

 * fixed a bug that annotation names are drawn for all row slices.

 * construct a valid path under Windows

 =======================

 CHANGES in VERSION 1.13.2
 
 * add `packLegend()`
 
 ========================
 

 CHANGES in VERSION 1.13.1
 
 * `Heatmap()`: add `km_title` to set the format of row title when `km` is set

 * `anno_link()`: add `extend` to extend the regions for the labels

 * `anno_boxplot()`: for row annotation, outliers are in the correct in y-axis.
   Thanks @gtg602c for the fix

 * `HeatmapAnnotation()`: gaps are included in the size of the annotations

 * `anno_link()`: graphic parameters are correctly reordered

 * `densityHeatmap()`: viewport is created with `clip = TRUE`

 * `decorate_*()`: add `envir` option to control where to look for variables inside `code`

 * `Legend()`: title supports expression

 * `anno_*()`: if the input is a data frame, warn that users may convert it to matrix

 ========================
 

 CHANGES in VERSION 1.11.8
 
 * `anno_barplot()`: accept a matrix as input to plot stacked barplots
 
 =========================
 

 CHANGES in VERSION 1.11.7
 
 * SingleAnnotation: if `col` is a vector with no names, it will be assigned
   as `level(value)` or `unique(value)`

 * HeatmapAnnotation: give warnings if color is defined while with no annotations

 * HeatmapAnnotation: check `col`, if it is not valid, give warnings

 * catch error when making annotation graphics

 
 ==========================
 

 CHANGES in VERSION 1.11.6
 
 * simply bump the verison number

 * `gap` in `Heatmap()` now can be a vector
 
 ===========================

 CHANGES in VERSION 1.11.5
 
 * `gap` in `HeatmapAnnotation` has been adjusted

 * annotations support drawing names of either sides

 * `densityHeatmap()`: quantile lines are also reordered

 * export `anno_oncoprint_barplot`

 * `Heatmap()`: if `col` is a unnamed vector and the number of colors is same as
   unique itemes in `mat`, the name of `col` vector is set to `sort(unique(mat))

 * adjusted the order of annotation legends
 * discreat legend: if a level is not in the data while defined by `col`, it will
   be removed.

 
 =============================
 

 CHANGES in VERSION 1.11.2
 

 * `grid.dendrogram()`: do not draw dendrogram if the height is zero

 * `densityHeatmap()`: support clustering on columns and more controls on column settings
 
 =============================
 

 CHANGES in VERSION 1.11.1
 
 * `draw,HeatmapList-method` can control row order and clustering of the main heatmap
 
 =============================
 

 CHANGES in VERSION 1.9.7
 
 * add `Legend()` function which is more flexible to generate
   different types of legends.
 
 ============================
 

 CHANGES in VERSION 1.9.6
 
 * `color_mapping_legend()`, there are ticks on continuous color bar
 
 =============================
 

 CHANGES in VERSION 1.9.5
 
 * add a section in the vignette to show how to adjust positions
   of column names when there are bottom annotations.

 * fixed a bug that character NA values can not to assigned with na_col

 * extra character 'at' and 'labels' in legends will be removed automatically

 * all arguments which are passed to `make_layout()` are all explicitly put
   in `draw()` instead of using ...

 
 =============================
 

 CHANGES in VERSION 1.9.4
 
 * heatmap bodied can be set as raster images if number of rows are too huge
 
 ==============================
 

 CHANGES in VERSION 1.9.3

 
 * graphic parameters are correctly recycled in row annotations

 * if there is only one row after splitting, there will be no dendrogram

 * add `range` option in `densityHeatmap()`

 * when `gap` is set for the main heatmap, other heatmps also adjust their `gap` values to it

 * fixed a bug that when rownames/colnames are not complete, dendrograms are corrupted

 * `alter_fun` now supports adding graphics grid by grid

 * add `show_pct` option in `oncoPrint()`

 * add `column_order` in `densityHeatmap()`

 
 ==============================
 

 CHANGES in VERSION 1.9.2

 
 * imporved example of `anno_link()`
 
 ==============================
 

 CHANGES in VERSION 1.9.1

 
 * width of the heatmap body are calculated correctly if it is set as a fixed unit
 * there is no dendrogram is nrows in a row-slice is 1
 * add `anno_link()` annotation function

 * bottom annotations are attached to the bottom edge of the heatmap 
   if there are additional blank space

 * colors for NA can be set by "_NA_" in annotations

 * `row_dend_reorder` and `column_dend_reorder` are set to `TRUE` by default again -_-!!
 * optimize the way to specify na_col in heatmap annotations
 * correct wrong viewport names in decorate_* functions

 
 ===============================
 

 CHANGES in VERSION 1.7.3
 
 * `oncoPrint()`: add `barplot_ignore` option to remove alterations
   that are not put on the barplot.

 * `oncoPrint()`: delete extra alter_fun if they are not in the matrix

 ================================
 

 CHANGES in VERSION 1.7.2
 
 * for `anno_points()`, `anno_barplot()`, `anno_boxplot()`, the name is assigned
   to the viewport with data ranges

 * oncoPrint(): top_annotation is assigned with correct height

 
 ===============================
 

 CHANGES in VERSION 1.7.1
 
 * `x` and `y` are `unit` object now in `cell_fun`
 * add an example to visualize GO game

 * transparency is kept when making discreate color mappings

 * `oncoPrint()`: barplots on top now are controlled by `top_annotation` argument

 
 ================================
 

 CHANGES in VERSION 1.5.1
 
 * `oncoPrint`: there are default graphics if type of alterations is 
   less than two.

 * `anno_*`: get rid of lazy loading

 
 =================================
 

 CHANGES in VERSION 1.4.4
 
 * NULL can be added to the heatmap list

 * give message if users mess up with row and column annotations

 * `oncoPrint`: columns are sorted by `memo sort` method to enhance
   the mutual exclusivity.

 
 ==================================
 

 CHANGES in VERSION 1.4.3
 
 * `anno_boxplot`: outliers can correctly plotted
 * add a new function `enhanced_basicplot`
 * fix a bug that order of annotation height is reversed
 
 ===================================
 

 CHANGES in VERSION 1.4.2
 

 * order of annotations can be adjusted arbitrarily

 * anno_barplot: scales are adjusted according to the baseline
 
 ===================================
 

 CHANGES in VERSION 1.4.1
 
 * revised the vignettes
 
 ===================================
 

 CHANGES in VERSION 1.4.0

 
 * returned value for `draw` method has been changes
 * add `row_order`, `column_order`, `row_dend` and `column_dend`
   to extract orders and dendrograms after heatmap clustering

 * add `select` function to interactively select sub region in the heatmap

   and retrieve row/column index in the selected sub region.

 * renames all varialbes `*hclust*` to `*dend*`

 
 ====================================
 

 CHANGES in VERSION 1.3.3
 
 * set `row_reorder` and `column_reorder` to FALSE by default in `Heatmap()`
 
 =====================================
 

 CHANGES in VERSION 1.3.2
 
 * fixed the error in the vignette header
 
 ======================================
 

 CHANGES in VERSION 1.3.1
 
 * padding of whole plot are correctly set now
 
 =======================================
 

 CHANGES in VERSION 1.3.0
 
 * parameters for legend are specified by a list
 * vignette was split into several vignettes which focus on specific topics
 * ColorMapping class suppoorts `at` and `labels`
 * add `row_anno_*` and `column_anno_*` functions
 * `anno_barplot` add `baseline` option to define the baseline of bars

 * change default colors for matrix

 * allow heatmap with zero-row
 * heatmap components can all have absolute units

 * add `oncoPriint()`

 
 =========================================
 

 CHANGES in VERSION 1.2.8
 
 * add decorate* functions
 
 =========================================
 

 CHANGES in VERSION 1.2.7
 
 * add `rows_reorder_weight` and `columns_reorder_weight` in `Heatmap()`
 * `draw,HeatmapList-method` returns a list of orders
 * clustering is stored as dendrogram internally
 
 =========================================
 

 CHANGES in VERSION 1.2.6
 
 * graphical parameters for row names can be set as same length
   as row slices
 * graphical parameters for annotation functions are consistent with
   the matrix now
 * insert `k = NULL` into anno_function if they are row annotations
 * add a new vignette "quick examples"

 * change style for continuous legend color bar

 * label order of discrete legend has been adjusted (now it is from top to bottom)

 
 =========================================
 

 CHANGES in VERSION 1.2.5
 
 * add dependency of base packages
 
 =========================================
 

 CHANGES in VERSION 1.2.4
 

 * name for ColorMapping object has default value now. 

 * legend for continuous values can be set as continuous legends
 * row title and column title as well as legend title support expression
 * add `heatmap_legend_title` in `Heatmap` and `ColorMapping`

 * fixed a bug in `plotDataFrame`

 
 ==========================================
 

 CHANGES in VERSION 1.2.3
 

 * `color_mapping_legend` can produce a continuous color legend

 * `color_mapping_legend` now returns a grob object
 
 ===========================================
 

 CHANGES in VERSION 1.2.2
 

 * adjust positions for row titles

 * rows can be split if `cluster_rows` are a clustering object
 * `row_order` and `column_order` can be set by dimension names
 
 ===========================================
 

 CHANGES in VERSION 1.2.1
 
 * adjust orders of row slices
 * support text rotation for heatmap titles 
 
 ===========================================
 

 CHANGES in VERSION 1.2.0
 
 * correct row orders if km and row_order is set in `Heatmap`
 
 ===========================================
 

 CHANGES in VERSION 1.1.9
 

 * adjust names and size of each component

 * add `densityHeatmap` function which visualizes column distribution
   in a matrix

 * add `plotDataFrame` function which provides a quick way to 
   visualize a data frame
 
 ============================================
 

 CHANGES in VERSION 1.1.8
 
 * dendrograms are implemented by `grid.segments` 

 * setting padding for each component
 * adjust gaps between heatmaps/row annotations if the column of the 
   heatmap is zero.

 
 ============================================
 
 

 CHANGES IN VERSION 1.1.7
 
 * check consistency of matrix row names
 
 =============================================
 

 CHANGES IN VERSION 1.1.6

 * improved generation of random colors
 * order of heatmap annotation legends are same as the order of annotations

 * support NA value in simple annotations

 * add examples in the vignette
 * simple row annotations now have the correct order
 * `col` can be a vector of colors if matrix is continuous
 * simple annotation can be logical

 ==============================================
 
 CHANGES IN VERSION 1.1.5
 
 * add gaps between heatmap components
 

 ===============================================
 

 CHANGES IN VERSION 1.1.4
 
 * support color mapping for NA values
 
 ===============================================
 

 CHANGES IN VERSION 1.1.3
 
 * Improved vignette to describe how to use row index in row annotation function
   if row annotations are also splitted by rows.
 
 ===============================================
 

 CHANGES IN VERSION 1.1.2
 
 * `anno_density`: graphics are now in correct order
 * add `anno_text`
 * add new examples corresponding to new functions
 
 ================================================
 

 CHANGES IN VERSION 1.1.1

 
 * fixed a bug when setting `cluster_rows` to FALSE but still cluster
   on rows.

 * add `rowAnnotation` and `columnAnnotation` functions
 * add examples in the vignette
 * No error if the heatmap list only contains zero-column matrix
 
 =================================================

 CHANGES IN VERSION 0.99.2
 

 * add two examples in vignette

 * add chunk labels in the vignette
 
 =================================================
 

 CHANGES IN VERSION 0.99.1
 
 * x and y in `cell_fun` are now `unit` objects.
 
 =================================================
 CHANGES IN VERSION 0.99.0

 
 * First release