plot_heatmap Error: cannot take a sample larger than the population when 'replace = FALSE'

[bednarsky]

Error:

Attaching package: ‘fastcluster’

The following object is masked from ‘package:stats’:

hclust

Error in sample.int(length(x), size, replace, prob) :
cannot take a sample larger than the population when 'replace = FALSE'
Calls: sample -> sample.int

• I don't get more info than this, so not even sure which sample call causes the issue
• input files seem fine
• potential clue: https://stackoverflow.com/questions/19648238/r-says-cannot-take-a-sample-larger-than-the-population-but-i-am-not-taking
• config settings:

  heatmap:
      metrics: ['correlation','cosine']
      hclust_methods: ['complete']
      n_observations: 1 # random sampled proportion float (0-1] or absolute number as integer
      n_features: 0.5 # highly variable features proportion float (0-1] or absolute number as integer

[bednarsky]

• this causes the error for me:

  unsupervised_analysis/workflow/scripts/plot_heatmap.R

  Line 127 in f902eff

  colors <- sample(unique(unlist(mapply(brewer.pal, qual_col_pals$maxcolors, rownames(qual_col_pals)))),n_cat)

• proposed fix:

+ randomColor <- function() {
+   paste0("#", paste0(sample(c(0:9, letters[1:6]), size = 6, replace = TRUE), collapse = ""))
+ }

# make color mapping
if (is.numeric(metadata[[metadata_col]])){
    #check if divergent or sequential?
    row_annot <- HeatmapAnnotation(df=metadata[,metadata_col, drop=FALSE], which="row")
    
}else{
    n_cat <- length(unique(metadata[[metadata_col]]))
    qual_col_pals = brewer.pal.info[brewer.pal.info$category == 'qual',]
    all_brewer_colors <- unique(unlist(mapply(brewer.pal, qual_col_pals$maxcolors, rownames(qual_col_pals))))
+    if (n_cat <= length(all_brewer_colors)){
+        # get nice colors
        colors <- sample(all_brewer_colors,n_cat)
+    }else{
+        # too many groups, get random colors
+        colors <- replicate(n_cat, randomColor())
+    }
    names(colors) <- unique(metadata[[metadata_col]])
    colors_list <- list()
    colors_list[[metadata_col]] <- colors# put here the mapped colors
    
    row_annot <- HeatmapAnnotation(df=metadata[,metadata_col, drop=FALSE], which="row", col=colors_list)
}

[sreichl]

Wow thats a lot of code instead of one line ^^. Sure this is all necessary?

If you have tested the proposed fix feel invited to make a PR.

[bednarsky]

It's not really that much more, highlighted it above now. Could also just remove the 'nice colors' part completely, but think it's nice that you first select for varying colors if there aren't that many

[sreichl]

ah okay, thought it was replacing only the one line haha

[bednarsky]

No, I just assumed you know all your code by heart :D