如何修改R中gbm_pheatmap包装函数的绘图参数（解决注释重叠）

Got it, let's tackle that overlapping annotation text issue in gbm_pheatmap. Since the function doesn't expose fontsize parameters for the annotations, we have a few straightforward ways to adjust this behavior:

Option 1: Modify the existing gbm_pheatmap function #

You can directly tweak the original function to add support for annotation fontsize controls. Here's how:

1. Start with the source code of gbm_pheatmap (you already have this, but you can re-fetch it anytime with print(gbm_pheatmap) in R).
2. Add fontsize-related parameters to the function definition, then pass them to the underlying pheatmap() call at the end.

Here's the modified function with added fontsize controls:

gbm_pheatmap <- function (gbm, genes_to_plot, cells_to_plot, n_genes = 5, colour = NULL, limits = c(-3, 3),
                          annotation_fontsize = 8, fontsize = 10) { # New fontsize parameters added
  if (!is.list(genes_to_plot)) {
    cat("Plotting one gene set instead of multiple cluster-specific gene sets\n")
    gene_indices <- sapply(genes_to_plot, function(x) get_gene_index(gbm, x))
    gene_annotation <- NULL
  } else {
    if ("significant" %in% names(genes_to_plot[[1]])) {
      gene_indices <- unlist(lapply(genes_to_plot, function(x) with(x, head(ix[significant], n_genes))))
      gene_grouping <- unlist(lapply(names(genes_to_plot), function(nm) rep(nm, with(genes_to_plot[[nm]], length(head(ix[significant], n_genes))))))
    } else {
      gene_indices <- unlist(lapply(genes_to_plot, function(x) x$ix[1:n_genes]))
      gene_grouping <- rep(names(genes_to_plot), each = n_genes)
    }
    gene_annotation <- data.frame(ClusterID = as.factor(gene_grouping))
  }
  cell_indices <- unlist(lapply(cells_to_plot, function(x) x$ix))
  value <- t(scale(t(as.matrix(exprs(gbm))[gene_indices, cell_indices])))
  value[value < limits[1]] <- limits[1]
  value[value > limits[2]] <- limits[2]
  rownames(value) <- make.unique(fData(gbm)$symbol[gene_indices])
  cell_grouping <- unlist(lapply(1:length(cells_to_plot), function(x) {
    rep(names(cells_to_plot)[x], length(cells_to_plot[[x]]$barcode))
  }))
  cell_annotation <- data.frame(ClusterID = as.factor(cell_grouping))
  rownames(cell_annotation) <- colnames(value)
  if (!is.null(gene_annotation)) {
    rownames(gene_annotation) <- rownames(value)
  }
  if (is.null(colour)) {
    anno_colors <- NULL
  } else {
    names(colour) <- names(cells_to_plot)
    anno_colors <- list(ClusterID = colour)
  }
  # Pass new fontsize parameters to the pheatmap call
  pheatmap(value, cluster_rows = FALSE, cluster_cols = FALSE, show_colnames = FALSE,
           annotation_row = gene_annotation, annotation_col = cell_annotation,
           annotation_names_row = FALSE, annotation_names_col = FALSE,
           annotation_colors = anno_colors,
           annotation_fontsize = annotation_fontsize, # Added parameter
           fontsize = fontsize) # Added parameter
}

Now when you call gbm_pheatmap(), you can adjust annotation_fontsize to a smaller value (like annotation_fontsize = 6) to fix the text overlap.

Option 2: Create a custom wrapper function #

If you don't want to overwrite the original gbm_pheatmap, just rename the modified function above to something like custom_gbm_pheatmap. This keeps the original function intact while letting you use your adjusted version for plotting.

Option 3: Temporarily adjust pheatmap's global defaults #

For a quick fix without modifying the function, you can set global defaults for pheatmap before calling gbm_pheatmap:

# Set default annotation fontsize to a smaller value
options(pheatmap.defaults = list(annotation_fontsize = 8))

# Run your gbm_pheatmap call as usual
gbm_pheatmap(...)

# Optional: Reset defaults after plotting if needed
options(pheatmap.defaults = NULL)

This works because pheatmap uses these global defaults when parameters aren't explicitly passed in the function call.

内容的提问来源于stack exchange，提问作者Atakan