Dev

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: immunogenetr
 Title: A Comprehensive Toolkit for Clinical HLA Informatics
-Version: 0.1.0
+Version: 0.2.0
 Authors@R: c(
     person("Nicholas", "Brown", , "nicholas.brown@pennmedicine.upenn.edu", role = c("cre", "aut"),
            comment = c(ORCID = "0000-0002-0046-2315")),

NEWS.md

@@ -1,3 +1,7 @@
+# immunogenetr 0.2.0
+
+* Updated `HLA_prefix_add` and `HLA_prefix_remove` to work on all alleles in a GL string. Also added the option of keeping locus designations in `HLA_prefix_remove`.
+
 # immunogenetr 0.1.0
 
 * Initial CRAN submission.

R/HLA_prefix_add.R

@@ -1,15 +1,15 @@
 #' @title HLA_prefix_add
 #'
 #' @description This function adds a specified prefix to the beginning of
-#' each value in the identified columns of the given data frame. Useful for adding
-#' HLA or gene prefixes.
+#' each HLA type, and works on a single allele or all alleles in a GL string.
+#' Useful for adding HLA or gene prefixes.
 #'
-#' @param data A string with a single HLA allele.
-#' @param prefix A character string to be added as a prefix to the column values.
+#' @param data A string with a single HLA allele, a GL string of HLA alleles,
+#' or a character vector containing either of the previous.
+#' @param prefix A character string to be added as a prefix to the alleles.
 #' Default is "HLA-".
 #'
-#' @return A data frame with the specified prefix added to the values in the
-#' selected columns.
+#' @return A vector with the specified prefix added to the values.
 #'
 #' @examples
 #' # The HLA_typing_LIS dataset contains a table as might be found in a clinical
@@ -28,8 +28,15 @@
 #' @export
 #'
 #' @importFrom stringr str_replace
+#' @importFrom dplyr mutate
+#' @importFrom dplyr %>%
 
 HLA_prefix_add <- function(data, prefix = "HLA-") {
-  # Add string to beginning of typing
-  str_replace(data, "^", prefix)
+  # Expand the allele of GL string
+  data %>%
+    GLstring_expand_longer() %>%
+    # Add string to beginning of typing
+    mutate(value = str_replace(value, "^", prefix)) %>%
+    # Collapse the GL string
+    ambiguity_table_to_GLstring()
 }

R/HLA_prefix_remove.R

@@ -1,11 +1,15 @@
 #' @title HLA_prefix_remove
 #'
-#' @description This function removes HLA and locus prefixes from a string of HLA typing:
-#' "HLA-A2" changes to "2".
+#' @description This function removes HLA and optionally locus prefixes from a string of HLA typing:
+#' "HLA-A2" changes to "A2" or "2". By default, HLA and locus prefixes are removed. This function
+#' also works on each allele in a GL string.
 #'
-#' @param data A string with a single HLA allele.
+#' @param data A string with a single HLA allele, a GL string of HLA alleles,
+#' or a character vector containing either of the previous.
+#' @param keep_locus A logical value indicating whether to retain any locus values.
+#' The default value is FALSE.
 #'
-#' @return A string modified to remove HLA and locus prefixes.
+#' @return A vector modified to remove HLA and optionally locus prefixes.
 #'
 #' @examples
 #' # The HLA_typing_1 dataset contains a table with HLA typing spread across multiple columns:
@@ -24,17 +28,29 @@
 #' @export
 #'
 #' @importFrom dplyr %>%
+#' @importFrom dplyr mutate
 #' @importFrom stringr str_replace
 
-HLA_prefix_remove <- function(data) {
+HLA_prefix_remove <- function(data, keep_locus = FALSE) {
   # Removes any HLA and locus prefixes from typing results.
-  data %>%
+  step_1 <- data %>%
+    GLstring_expand_longer() %>%
     # replaces "HLA-" with an empty string in each cell.
-    str_replace("HLA-", "") %>%
-    # replaces any sequences of alphabetic characters at the start of the string with an empty string
-    str_replace("^[:alpha:]+", "") %>%
-    # replaces any sequences of digits followed by an asterisk with an empty string
-    str_replace(., "[:digit:]*\\*", "")
+    mutate(value = str_replace(value, "HLA-", ""))
+
+  if (keep_locus) {
+    step_2 <- step_1
+  } else {
+    # Remove the locus.
+    step_2 <- step_1 %>%
+      # replaces any sequences of alphabetic characters at the start of the string with an empty string
+      mutate(value = str_replace(value, "^[:alpha:]+", "")) %>%
+      # replaces any sequences of digits followed by an asterisk with an empty string
+      mutate(value = str_replace(value, "[:digit:]*\\*", ""))
+  }
+
+  # Reassemble the GL string
+  ambiguity_table_to_GLstring(step_2)
 }
 
-globalVariables(c("."))
+globalVariables(c("step_1", "step_2"))

R/HLA_validate.R

@@ -10,7 +10,7 @@
 #'
 #' @param data A string containing an HLA allele.
 #'
-#' @return A string with a valid HLA allele.
+#' @return A string with a valid HLA allele or NA if no valid allele was present.
 #'
 #' @examples
 #' HLA_validate("HLA-A2")

cran-comments.md

@@ -2,4 +2,4 @@
 
 0 errors ✔ | 0 warnings ✔ | 0 notes ✔
 
-* This is a new release.
+* Updated `HLA_prefix_add` and `HLA_prefix_remove` to work on all alleles in a GL string. Also added the option of keeping locus designations in `HLA_prefix_remove`.

immunogenetr.Rproj

@@ -1,4 +1,5 @@
 Version: 1.0
+ProjectId: 4b2ac85c-cc00-4118-bdf2-a85d3de96ab2
 
 RestoreWorkspace: Default
 SaveWorkspace: Default

tests/testthat/test-HLA_prefix_add.R

@@ -25,4 +25,10 @@ test_that("HLA_prefix_add correctly adds prefixes", {
 
   result_special_chars <- HLA_prefix_add("**01:01", "HLA-")
   expect_equal(result_special_chars, "HLA-**01:01")
+
+  result_gl_string <- HLA_prefix_add("A*02:01/A*02:02+A*68:01^B*57:01+B*07:02", "HLA-")
+  expect_equal(result_gl_string, "HLA-A*02:01/HLA-A*02:02+HLA-A*68:01^HLA-B*57:01+HLA-B*07:02")
+
+  result_gl_string_vector <- HLA_prefix_add(c("A*02:01/A*02:02+A*68:01^B*57:01+B*07:02", "A*02:01", "B*07:01/B*07:02+B*15:01"), "HLA-")
+  expect_equal(result_gl_string_vector, c("HLA-A*02:01/HLA-A*02:02+HLA-A*68:01^HLA-B*57:01+HLA-B*07:02", "HLA-A*02:01", "HLA-B*07:01/HLA-B*07:02+HLA-B*15:01"))
 })

tests/testthat/test-HLA_prefix_remove.R

@@ -25,4 +25,10 @@ test_that("HLA_prefix_remove correctly removes prefixes", {
 
   result_special_chars <- HLA_prefix_remove("HLA-**01:01")
   expect_equal(result_special_chars, "*01:01")
+
+  result_GL_string <- HLA_prefix_remove(c("HLA-A*02:01", "HLA-A2", "HLA-A*02:01/HLA-A*02:02+HLA-A*68:01^HLA-B*57:01+HLA-B*07:02"), keep_locus = TRUE)
+  expect_equal(result_GL_string, c("A*02:01", "A2", "A*02:01/A*02:02+A*68:01^B*57:01+B*07:02"))
+
+  result_GL_string_2 <- HLA_prefix_remove(c("HLA-A*02:01", "HLA-A2", "HLA-A*02:01/HLA-A*02:02+HLA-A*68:01^HLA-B*57:01+HLA-B*07:02"), keep_locus = FALSE)
+  expect_equal(result_GL_string_2, c("02:01", "2", "02:01/02:02+68:01^57:01+07:02"))
 })