Improve calculate go enrichment

NAMESPACE

@@ -147,6 +147,7 @@ importFrom(rlang,ensym)
 importFrom(rlang,expr)
 importFrom(rlang,new_formula)
 importFrom(rlang,sym)
+importFrom(scales,number_format)
 importFrom(stats,median)
 importFrom(stats,na.omit)
 importFrom(stats,p.adjust)

NEWS.md

@@ -1,8 +1,15 @@
-# protti 0.8.9000
+# protti 0.8.0.9000
+
+## New features 
+
+* `calculate_go_enrichment()` got additional arguments.
+  * `replace_long_name`: a logical argument that specifies if GO term names above 50 characters should be replaced by the GO ID instead for the plot. This ensures that the plotting area doesn't become too small due to the long name. The default is `TRUE`.
+  * `label_move_frac`: a numeric argument between 0 and 1 that specifies which labels should be moved outside of the bar. The default is 0.2, which means that the labels of all bars that have a size of 20% or less of the largest bar are moved to the right of the bar. This prevents labels from  overlapping with the bar boundaries.
 
 ## Bug fixes
 
 * `fetch_uniprot()` previously had an issue where it incorrectly identified certain IDs as UniProt IDs, such as ENSEMBL IDs. For example, it would incorrectly interpret `"CON_ENSEMBL:ENSBTAP00000037665"` as `"P00000"`. To address this, the function now requires that UniProt IDs are not preceded or followed by letters or digits. This means that UniProt IDs should be recognized only if they stand alone or are separated by non-alphanumeric characters. For instance, in the string `"P02545;P20700"`, both `"P02545"` and `"P20700"` are correctly identified as UniProt IDs because they are separated by a semicolon and not attached to any other letters or digits.
+* `calculate_go_enrichment()` now correctly uses to total number of provided proteins for the contingency table. Previously it falsely only considered proteins with a GO annotation for the enrichment analysis.
 
 # protti 0.8.0
 

R/calculate_go_enrichment.R

@@ -82,6 +82,13 @@ go_enrichment <- function(...) {
 #' determines if the enrichment analysis should be performed in order to check for both enrichemnt and
 #' deenrichemnt or only one of the two. This affects the statistics performed and therefore also the displayed
 #' plot.
+#' @param replace_long_name a logical argument that specifies if GO term names above 50 characters should
+#' be replaced by the GO ID instead for the plot. This ensures that the plotting area doesn't become
+#' too small due to the long name. The default is `TRUE`.
+#' @param label_move_frac a numeric argument between 0 and 1 that specifies which labels should be
+#' moved outside of the bar. The default is 0.2, which means that the labels of all bars that have a size
+#' of 20% or less of the largest bar are moved to the right of the bar. This prevents labels from
+#' overlapping with the bar boundaries.
 #' @param min_n_detected_proteins_in_process is a numeric argument that specifies the minimum number of
 #' detected proteins required for a GO term to be displayed in the plot. The default is 1, meaning
 #' no filtering of the plotted data is performed. This argument does not affect any computations or
@@ -111,6 +118,7 @@ go_enrichment <- function(...) {
 #' @importFrom rlang .data !! ensym
 #' @importFrom magrittr %>%
 #' @importFrom purrr map
+#' @importFrom scales number_format
 #' @export
 #'
 #' @examples
@@ -217,6 +225,8 @@ calculate_go_enrichment <- function(data,
                                     heatmap_fill_colour_rev = TRUE,
                                     label = TRUE,
                                     enrichment_type = "all",
+                                    replace_long_name = TRUE,
+                                    label_move_frac = 0.2,
                                     min_n_detected_proteins_in_process = 1,
                                     plot_cutoff = "adj_pval top10") {
   # to avoid note about no global variable binding. Usually this can be avoided with
@@ -331,7 +341,6 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
 
   # group argument is not missing
   cont_table <- go_data %>%
-    tidyr::drop_na(.data$go_id, {{ is_significant }}) %>%
     { # group argument is missing
       if (group_missing) {
         dplyr::group_by(., {{ is_significant }})
@@ -355,7 +364,8 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
       }
     } %>%
     tidyr::complete(.data$go_id, tidyr::nesting(!!rlang::ensym(is_significant), n_sig), fill = list(n_has_process = 0)) %>%
-    dplyr::ungroup()
+    dplyr::ungroup() %>%
+    tidyr::drop_na(.data$go_id)
 
 
   if (group_missing) {
@@ -438,6 +448,11 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
     dplyr::ungroup() %>%
     dplyr::filter(.data$n_detected_proteins_in_process >= min_n_detected_proteins_in_process)
 
+  if (replace_long_name) {
+    filtered_result_table <- filtered_result_table %>%
+      mutate(term = ifelse(nchar(.data$term) > 50, .data$go_id, .data$term))
+  }
+
   if (!missing(group) & y_axis_free & plot_style == "barplot") {
     # arrange table by group and go term for plot
     # this ensures that the terms are in the right order for a facet plot with a free axis
@@ -449,11 +464,13 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
   if (stringr::str_detect(plot_cutoff, pattern = "top")) {
     split_cutoff <- stringr::str_split(plot_cutoff, pattern = " ", simplify = TRUE)
     type <- split_cutoff[1]
-    top <- stringr::str_extract(split_cutoff[2], pattern = "\\d+")
+    top <- as.numeric(stringr::str_extract(split_cutoff[2], pattern = "\\d+"))
     plot_input <- filtered_result_table %>%
       dplyr::ungroup() %>%
       dplyr::mutate(neg_log_sig = -log10(!!rlang::ensym(type))) %>%
-      dplyr::slice(1:top)
+      dplyr::group_by({{ group }}) %>%
+      dplyr::mutate(n = 1:dplyr::n()) %>%
+      dplyr::filter(n <= top)
   } else {
     split_cutoff <- stringr::str_split(plot_cutoff, pattern = " ", simplify = TRUE)
     type <- split_cutoff[1]
@@ -464,6 +481,10 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
       dplyr::filter(!!rlang::ensym(type) <= threshold)
   }
 
+  # move label if bar is less than 20% (default) of largest bar
+  plot_input <- plot_input %>%
+    mutate(hjust = ifelse((.data$neg_log_sig / max(.data$neg_log_sig)) < label_move_frac, -0.15, 1.05))
+
   if (plot_style == "barplot") {
     # Check if ggforce package is available. If not prompt user to install it.
     if (!requireNamespace("ggforce", quietly = TRUE)) {
@@ -509,13 +530,13 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
                 "%)"
               ),
               y = .data$neg_log_sig - 0.1,
-              hjust = 1
+              hjust = .data$hjust
             )
           )
         }
       } +
       ggplot2::scale_fill_manual(values = c(Deenriched = barplot_fill_colour[1], Enriched = barplot_fill_colour[2])) +
-      ggplot2::scale_y_continuous(breaks = seq(0, 100, 1)) +
+      ggplot2::scale_y_continuous(labels = scales::number_format(accuracy = 1)) +
       ggplot2::coord_flip() +
       {
         if (!missing(group)) {