Skip to contents

Plot pretty heatmap with annotations for co-dilution data

Source: R/plot-pheatmap.R
pheatmap_with_anno_cd.Rd

Plot pretty heatmap with annotations for co-dilution data

Usage

pheatmap_with_anno_cd(
  dt_metrics,
  normalization_type = "GR",
  metric = "xc50",
  fit_source = "gDR",
  hm_title = NA,
  colors_vec = NULL,
  no_breaks = 50,
  cluster_rows = TRUE,
  cluster_cols = TRUE,
  distfun = compute_distances,
  annotation_row = NULL,
  annotation_col = NULL,
  annotation_colors = NULL
)

Arguments

dt_metrics

data.table representing data from the Metrics assay, outputted by gDRutils::convert_se_assay_to_dt and co-dilution SummarizedExperiment

normalization_type

string with normalization types to be selected one of: "GR" ("GRvalue") or "RV" ("RelativeViability")

metric

string name of the metric; one of: "xc50" ("GR50" or "IC50" - respectively depending on normalization_type), "x_max" ("GR Max" or "E Max") or "x_mean" ("GR Mean" or "RV Mean")

fit_source

string source name for metrics

hm_title

string plot title

colors_vec

character vector of colors (valid name or hex) used in heatmap; note that the first color will be assigned to the min value, and the last one - to the max

no_breaks

numeric number of breaks on scale used for mapping values to colors

cluster_rows

logical flag whether indicating rows should be clustered; the dendrogram will not be shown for the matrix with any dimension greater than 200.

cluster_cols

logical flag indicating whether columns should be clustered; the dendrogram will not be shown for the matrix with any dimension greater than 200.

distfun

function used to compute the distance (dissimilarity) between both rows and columns; used for the dendrogram when cluster_rows or cluster_cols is set to TRUE the default is compute_distances using Spearman method.

annotation_row

data.table that specifies the annotations shown on the left side of the heatmap. Each row defines the features for a specific row. The rows in the data and in the annotation are matched using corresponding names from the required DrugName, DrugName_2 and Concentration_2 columns. Note that color schemes takes into account if the variable is continuous or discrete.

annotation_col

data.table that specifies the annotations shown above the heatmap. Each row defines the features for a specific column. The columns in the data and in the annotation are matched using corresponding names from the required CellLineName column. Note that color schemes takes into account if the variable is continuous or discrete.

annotation_colors

named list for specifying annotation_col and annotation_row track colors manually; note list is named with annotation name (column names of annotation_row - without DrugName and column names of annotation_col - without CellLineName), each list item is named vector with valid color name for each value described in annotation_row and in annotation_col, respectively. Not described elements will be colored by default.

Value

A named list with elements:

  • data a list containing the information visualized in the heatmap:

    • matrix data shown in the heatmap for the selected metric.

    • annotation_row a table with row annotations (for DrugName), if provided.

    • annotation_col a table with column annotations (for CellLineName), if provided.

  • pheatmap object containing the heatmap itself.

See also

pheatmap::pheatmap

Author

Janina Smoła janina.smola@external.roche.com

Examples

mae <- gDRutils::get_synthetic_data("combo_codilution_small")
se <- mae[[gDRutils::get_supported_experiments("cd")]]
dt_metrics <- gDRutils::convert_se_assay_to_dt(se = se,
                                               assay_name = "Metrics")

output <- pheatmap_with_anno_cd(dt_metrics = dt_metrics)
hm_1 <- output[["heatmap"]]
patchwork::wrap_elements(hm_1[["gtable"]])


annotation_manual_col <-
  unique(dt_metrics[, c("CellLineName", "Tissue"), with = FALSE])
annotation_manual_row <-
  unique(dt_metrics[, c("DrugName", "DrugName_2", "Concentration_2",
                        "drug_moa", "drug_moa_2"),
                    with = FALSE])
annotation_map <-
  get_ann_color_map(unique(dt_metrics[, c("Tissue", "drug_moa", "drug_moa_2"), with = FALSE]))

output <- pheatmap_with_anno_cd(dt_metrics = dt_metrics,
                                normalization_type = "RV",
                                metric = "x_mean",
                                colors_vec = c("darkblue", "grey90"),
                                annotation_row = annotation_manual_row,
                                annotation_col = annotation_manual_col,
                                annotation_colors = annotation_map)
hm_2 <- output[["heatmap"]]
patchwork::wrap_elements(hm_2[["gtable"]])


annotation_manual <- data.table::data.table(
  CellLineName =
    c("cellline_AA", "cellline_EA", "cellline_IB", "cellline_MC", "cellline_BC"),
  mut_A = c(1, 1, 1, 0, 0),
  mut_B = c("yes", "yes", "no", "no", "no")
)
annotation_map <- list(
  mut_A = c("1" = "coral", "0" = "cadetblue"),
  mut_B = c("yes" = "black", "no" = "grey90")
)

output <- pheatmap_with_anno_cd(dt_metrics = dt_metrics,
                                annotation_col = annotation_manual,
                                annotation_colors = annotation_map,
                                hm_title = get_hm_title(
                                  normalization_type = "GR",
                                  metric = "hsa_score",
                                  dataset_name = "Co-dilution data"))
hm_3 <- output[["heatmap"]]
patchwork::wrap_elements(hm_3[["gtable"]])