A memory bug fix

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: fenr
 Title: Fast functional enrichment for interactive applications
-Version: 1.3.1
+Version: 1.5.1
 Authors@R: person(
     given = "Marek",
     family = "Gierlinski",
@@ -53,6 +53,6 @@ Suggests:
     knitr,
     rmarkdown,
     topGO
-RoxygenNote: 7.3.1
+RoxygenNote: 7.3.2
 VignetteBuilder: knitr
 LazyData: false

NEWS.md

@@ -171,3 +171,6 @@
 
  - Go term namespace added to the information extracted by `fetch_go`.
 
+## Version 1.4.1
+
+ - Attempted to fix a bizarre error message on Bioconductor's test machines with older version of MacOS. Windows and Linux are not affected; my laptop running Sequoia 5.2 does not show show errors. I suspect a memory leak in older systems. The error `vector memory limit of 64.0 Gb reached, see mem.maxVSize()` happened in the function parse_kegg_genes(), a flat-file parser for KEGG. It occurred around the call tidyr::separate(), which I replaced with an alternative approach. Will see if the error is fixed.

R/kegg.R

@@ -96,16 +96,20 @@ parse_kegg_genes <- function(s) {
       i <- i + 1
     }
 
-    # create final tibble, attempt to extract gene symbols when semicolon is found
-    genes |>
-      tibble::as_tibble_col(column_name = "data") |>
-      tidyr::separate(data, c("gene_id", "gene_symbol"), sep = "\\s+", extra = "merge") |>
-      dplyr::mutate(gene_symbol = dplyr::if_else(
-        stringr::str_detect(gene_symbol, ";"),
-        stringr::str_remove(gene_symbol, ";.+$"),
-        gene_id
-      )) |>
-      tibble::add_column(term_id = pathway)
+    purrr::map(genes, function(gene) {
+      # First element - gene ID, second - gene symbol, if contains semicolon
+      v <- stringr::str_split_1(gene, "\\s+")
+      tibble::tibble(
+        gene_id = v[1],
+        gene_symbol = ifelse(
+          stringr::str_detect(v[2], ";"),
+          stringr::str_remove(v[2], ";.*$"),
+          v[1]
+        ),
+        term_id = pathway
+      )
+    }) |> 
+      purrr::list_rbind()
   }) |>
     purrr::list_rbind()
 }