heatmap_column_order function

Return Heatmap column order from ComplexHeatmap heatmap object

heatmap_column_order(hm, which_heatmap = NULL)

Arguments

• hm: Heatmap or HeatmapList object as defined by the Bioconductor package via ComplexHeatmap::Heatmap().

• which_heatmap: used to specify a specific heatmap with hm is provided as a HeatmapList. When NULL (default) the first heatmap in hm@ht_list is used. When one value is supplied, only that heatmap is used. When multiple values are supplied, a list is returned. Input can be either:

  • numeric - indicating the heatmap number in hm@ht_list
  • character - indicating the heatmap name seen in names(hm@ht_list)

Returns

output depends upon the heatmap:

• When heatmap columns are grouped using column_split, and when the data matrix contains colnames, returns a character vector of colnames in the order they appear in the heatmap. When there are no colnames, integer

  column index values are returned. If the heatmap has column labels, they are returned as vector names.

• When columns are grouped using column_split, it returns a list of vectors as described above. The list

  is named using the column_title labels only when there is an equal number of column labels.

Details

This function is a helpful utility to return the fully qualified list of colnames in a ComplexHeatmap::Heatmap

object.

The core intention is for the output to be usable with the original data matrix used in the heatmap. Therefore, the vector values are colnames() when present, or integer

column index values when there are no colnames(). If heatmap column_labels are defined, they are returned as names().

Note that names() are assigned inside try() to allow the case where column_labels, or column_title labels cannot be coerced to character values, for example using gridtext

for markdown formatting.

Examples

if (check_pkg_installed("ComplexHeatmap")) {
   set.seed(123);

   mat <- matrix(stats::rnorm(18 * 24),
      ncol=24);
   rownames(mat) <- paste0("row", seq_len(18))
   colnames(mat) <- paste0("column", seq_len(24))

   # obtaining row order first causes a warning message
   hm1 <- ComplexHeatmap::Heatmap(mat);

   # best practice is to draw() and store output in an object
   # to ensure the row orders are absolutely fixed
   hm1_drawn <- ComplexHeatmap::draw(hm1);
   print(heatmap_row_order(hm1_drawn))
   print(heatmap_column_order(hm1_drawn))

   # row and column split
   hm1_split <- ComplexHeatmap::Heatmap(mat,
      column_split=3, row_split=3, border=TRUE);
   hm1_split_drawn <- ComplexHeatmap::draw(hm1_split);
   print(heatmap_row_order(hm1_split_drawn))
   print(heatmap_column_order(hm1_split_drawn))

   # display two heatmaps side-by-side
   mat2 <- mat + stats::rnorm(18*24);
   hm2 <- ComplexHeatmap::Heatmap(mat2, border=TRUE, row_split=4);

   hm1hm2_drawn <- ComplexHeatmap::draw(hm1_split + hm2,
      ht_gap=grid::unit(1, "cm"));
   print(heatmap_row_order(hm1hm2_drawn))
   print(heatmap_row_order(hm1hm2_drawn, which_heatmap=2))
   # by default the order uses the first heatmap
   print(heatmap_column_order(hm1hm2_drawn))
   # the second heatmap can be returned
   print(heatmap_column_order(hm1hm2_drawn, which_heatmap=2))
   # or a list of heatmap orders can be returned
   print(heatmap_column_order(hm1hm2_drawn, which_heatmap=1:2))

   # stacked vertical heatmaps
   hm1hm2_drawn_tall <- ComplexHeatmap::draw(
      ComplexHeatmap::`%v%`(hm1_split, hm2),
      ht_gap=grid::unit(1, "cm"));
   print(heatmap_row_order(hm1hm2_drawn))
   print(heatmap_row_order(hm1hm2_drawn, which_heatmap=2))
   print(heatmap_row_order(hm1hm2_drawn, which_heatmap=1:2))
   print(heatmap_row_order(hm1hm2_drawn,
      which_heatmap=names(hm1hm2_drawn@ht_list)))

   # annotation heatmap
   ha <- ComplexHeatmap::rowAnnotation(left=rownames(mat))
   ha_drawn <- ComplexHeatmap::draw(ha + hm1)
   print(sdim(ha_drawn@ht_list))
   print(heatmap_row_order(ha_drawn))
   print(heatmap_column_order(ha_drawn))

   # stacked vertical heatmaps with top annotation
   ta <- ComplexHeatmap::HeatmapAnnotation(top=colnames(mat))
   hm1_ha <- ComplexHeatmap::Heatmap(mat,
      left_annotation=ha,
      column_split=3, row_split=3, border=TRUE);
   hm1hm2_drawn_tall <- ComplexHeatmap::draw(
      ComplexHeatmap::`%v%`(ta,
         ComplexHeatmap::`%v%`(hm1_ha, hm2)),
      ht_gap=grid::unit(1, "cm"));
   print(sdim(hm1hm2_drawn_tall@ht_list))
   print(heatmap_row_order(hm1hm2_drawn_tall))
   print(heatmap_row_order(hm1hm2_drawn_tall, 2))
}

See Also

Other jam heatmap functions: cell_fun_label(), heatmap_row_order()