Features/extend aa score

NEWS.md

@@ -5,6 +5,7 @@
 * `calculate_peptide_abundance()` new function to calculate peptide abundances from precursor abundances.
 * `fit_drc_4p()` received the `show_progress` argument that is by default `TRUE` and allows the user to show or hide progress bars. This closes issue #278.
 * `assign_missingness()` gained the `completeness_MNAR_reference` argument, allowing users to control how complete a condition must be to be considered sufficiently observed when assigning MNAR missingness. This closes issue #200.
+* `calculate_aa_scores()` gained a `methods` argument, allowing user to choose between additive or the new multiplicative score calculation mode. The function now also returns min-max normalised scores in addition to raw values.
 
 ## Bug fixes
 

R/calculate_aa_scores.R

@@ -19,6 +19,8 @@
 #' data frame. Default is not retaining additional columns \code{retain_columns = NULL}. Specific
 #' columns can be retained by providing their names (not in quotations marks, just like other
 #' column names, but in a vector).
+#' @param method a character argument selecting the method used for score calculation.
+#' Supported are "multiplicative" = `-log10(adj_pval) * abs(diff)` (default) or "additive" = `-log10(adj_pval) + abs(diff)`.
 #'
 #' @return A data frame that contains the aggregated scores per amino acid position, enabling to
 #' draw fingerprints for each individual protein.
@@ -51,30 +53,57 @@ calculate_aa_scores <- function(data,
                                 adj_pval = adj_pval,
                                 start_position,
                                 end_position,
-                                retain_columns = NULL) {
+                                retain_columns = NULL,
+                                method = "multiplicative") {
+  # validate method input
+  method <- match.arg(method, c("multiplicative", "additive"))
+
   output <- data %>%
     dplyr::ungroup() %>%
     dplyr::distinct({{ protein }}, {{ diff }}, {{ adj_pval }}, {{ start_position }}, {{ end_position }}) %>%
     tidyr::drop_na({{ diff }}, {{ adj_pval }}) %>%
-    dplyr::mutate(score = -log10({{ adj_pval }}) * abs({{ diff }})) %>%
+    dplyr::mutate(
+      score = if (method == "multiplicative") {
+        -log10({{ adj_pval }}) * abs({{ diff }})
+      } else if (method == "additive") {
+        -log10({{ adj_pval }}) + abs({{ diff }})
+      }
+    ) %>%
     dplyr::rowwise() %>%
     dplyr::mutate(residue = list(seq({{ start_position }}, {{ end_position }}))) %>%
     tidyr::unnest("residue") %>%
     dplyr::group_by({{ protein }}, .data$residue) %>%
     dplyr::mutate(amino_acid_score = mean(.data$score)) %>%
     dplyr::distinct({{ protein }}, .data$residue, .data$amino_acid_score)
 
+  # normalization (per protein)
+  output <- output %>%
+    dplyr::group_by({{ protein }}) %>%
+    dplyr::mutate(
+      amino_acid_score_normalized = {
+        min_val <- min(.data$amino_acid_score, na.rm = TRUE)
+        max_val <- max(.data$amino_acid_score, na.rm = TRUE)
+        if (max_val == min_val) {
+          1 # avoid division by zero; constant protein gets 1
+        } else {
+          (.data$amino_acid_score - min_val) / (max_val - min_val)
+        }
+      }
+    ) %>%
+    dplyr::ungroup()
 
   if (!missing(retain_columns)) {
     output <- data %>%
       dplyr::select(!!enquo(retain_columns), colnames(output)[!colnames(output) %in% c(
         "residue",
-        "amino_acid_score"
+        "amino_acid_score",
+        "amino_acid_score_normalized"
       )]) %>%
       dplyr::distinct() %>%
       dplyr::right_join(output, by = colnames(output)[!colnames(output) %in% c(
         "residue",
-        "amino_acid_score"
+        "amino_acid_score",
+        "amino_acid_score_normalized"
       )])
   }
 

R/calculate_go_enrichment.R

@@ -385,7 +385,7 @@ if you used the right organism ID.", prefix = "\n", initial = ""))
     groups_to_skip <- cont_table %>%
       dplyr::group_by({{ group }}) %>%
       dplyr::summarise(n_levels = dplyr::n_distinct({{ is_significant }}), .groups = "drop") %>%
-      dplyr::filter(n_levels < 2) %>%
+      dplyr::filter(.data$n_levels < 2) %>%
       dplyr::pull({{ group }})
 
     cont_table <- cont_table %>%

R/qc_sample_correlation.R

@@ -136,7 +136,7 @@ qc_sample_correlation <- function(data,
     heatmap_interactive <-
       heatmaply::heatmaply(
         correlation,
-        main = "Correlation based hirachical clustering of samples",
+        main = "Correlation based hierarchical clustering of samples",
         col_side_colors = annotation,
         col_side_palette = c(
           annotation_colours[[1]],
@@ -190,7 +190,7 @@ qc_sample_correlation <- function(data,
         cluster_cols = stats::as.hclust(dendrogram_column),
         annotation = annotation,
         annotation_colors = annotation_colours,
-        main = "Correlation based hierachical clustering of samples",
+        main = "Correlation based hierarchical clustering of samples",
         color = viridis_colours,
         silent = TRUE
       )

tests/testthat/test-workflow.R

@@ -852,7 +852,7 @@ test_that("calculate_aa_scores works", {
 
   expect_is(aa_fingerprint, "data.frame")
   expect_equal(nrow(aa_fingerprint), 45)
-  expect_equal(ncol(aa_fingerprint), 3)
+  expect_equal(ncol(aa_fingerprint), 4)
 })
 
 # Test for random forest imputation