Use cpp11bigwig for read_bigwig() (#425)

* Use cpp11bigwig for `read_bigwig()`

* Update news

Author: jayhesselberth

DESCRIPTION

@@ -21,6 +21,7 @@ Depends:
 Imports:
     broom,
     cli,
+    cpp11bigwig,
     dplyr (>= 0.8.0),
     ggplot2,
     lifecycle,

NEWS.md

@@ -1,20 +1,23 @@
 # valr (development version)
 
-* `read_bigwig()` and `read_gtf()` were deprecated. The rtracklayer package used
-for this functionality is no longer a dependency of valr due to errors from 
+* `read_bigwig()` now uses cpp11bigwig on CRAN. The `set_strand` param was removed to be
+  more consistent with expected bigWig contents.
+
+* `read_gtf()` was deprecated. The rtracklayer package used
+for this functionality is no longer a dependency of valr due to errors from
 CRAN AddressSantizer checks of the UCSC c-library code vendored in rtracklayer.
 
 * valr now depends on R >= 4.0.0.
 
 # valr 0.8.2
 
-* Address NOTE on CRAN about Rd link targets.  
+* Address NOTE on CRAN about Rd link targets.
 
-* Change maintainer email address. 
+* Change maintainer email address.
 
 # valr 0.8.1
 
-* Make vdiffr dependency optional during package testing. 
+* Make vdiffr dependency optional during package testing.
 
 # valr 0.8.0
 
@@ -26,17 +29,17 @@ CRAN AddressSantizer checks of the UCSC c-library code vendored in rtracklayer.
 
 # valr 0.6.8
 
-* `bed_closest()` now reports all x intervals, even when there are no closest y intervals (e.g. when there is no matching chromosome in y intervals). These intervals are returned populated with `NA` for `.overlap`, `.dist` and y interval locations. 
+* `bed_closest()` now reports all x intervals, even when there are no closest y intervals (e.g. when there is no matching chromosome in y intervals). These intervals are returned populated with `NA` for `.overlap`, `.dist` and y interval locations.
 
-* Reimplemented `bed_closest()` to use binary search rather than an interval tree search. The closest y interval can be missed with the previous search strategy in high depth interval trees. 
+* Reimplemented `bed_closest()` to use binary search rather than an interval tree search. The closest y interval can be missed with the previous search strategy in high depth interval trees.
 
 * Fix off by one error when using `max_dist` argument in `bed_cluster()` (#401).
 
 # valr 0.6.7
 
 * Removed `SystemRequirements` from DESCRIPTION to eliminate a NOTE on CRAN.
 
-* `bed_coverage()` now reports intervals from `x` with no matching group in `y` (#395).  
+* `bed_coverage()` now reports intervals from `x` with no matching group in `y` (#395).
 
 # valr 0.6.6
 
@@ -49,15 +52,15 @@ CRAN AddressSantizer checks of the UCSC c-library code vendored in rtracklayer.
 
 # valr 0.6.5
 
-* Handle `max_dist` for first intervals in `bed_cluster()` (#388) 
+* Handle `max_dist` for first intervals in `bed_cluster()` (#388)
 
 # valr 0.6.4
 
-* Fixed intron score numbering error in `create_introns` (#377 @sheridar) 
+* Fixed intron score numbering error in `create_introns` (#377 @sheridar)
 
-* Fixed bug in handling of list inputs for `bed_intersect()`(#380 @sheridar)   
+* Fixed bug in handling of list inputs for `bed_intersect()`(#380 @sheridar)
 
-* Added `read_bigwig` and `read_gtf` functions to import data into valr compatible tibbles (#379)  
+* Added `read_bigwig` and `read_gtf` functions to import data into valr compatible tibbles (#379)
 
 * Kent Riemondy is now maintainer.
 
@@ -69,7 +72,7 @@ CRAN AddressSantizer checks of the UCSC c-library code vendored in rtracklayer.
 
 ## Minor changes
 
-* `RMariaDB` has replaced the deprecated `RMySQL` package as the database backend. 
+* `RMariaDB` has replaced the deprecated `RMySQL` package as the database backend.
 
 * valr now imports Rcpp, which should have always been the case,
 but was masked by its Import by readr, which recently dropped use of Rcpp.
@@ -84,17 +87,17 @@ but was masked by its Import by readr, which recently dropped use of Rcpp.
 
 ## Major changes
 
-* `trbl_interval()` and `trbl_genome()` custom `tibble` subclasses have been deemed unnecessary and have been removed from the package. 
+* `trbl_interval()` and `trbl_genome()` custom `tibble` subclasses have been deemed unnecessary and have been removed from the package.
 
-* coercing `GRanges` to a `valr` compatible data.frame now uses the `gr_to_bed()` function rather than `as.trbl_interal()` methods. 
+* coercing `GRanges` to a `valr` compatible data.frame now uses the `gr_to_bed()` function rather than `as.trbl_interal()` methods.
 
 ## Minor changes
 
 * dplyr version < 0.8.0 is no longer supported due to unnecessary code bloat and challenges with handling multiple grouping structures (#359).
 
 * The `sort_by` argument of `bed_random()` has been changed to `sorted`, and will now by default
 use `bed_sort()` to sort the output, rather than rely on naming the sorting columns. Sorting can
-be suppressed by using `sorted = FALSE`. 
+be suppressed by using `sorted = FALSE`.
 
 * `bed_sort()` now uses base R sorting with the `radix` method for increased speed. (#353)
 
@@ -108,13 +111,13 @@ be suppressed by using `sorted = FALSE`.
 
 # valr 0.5.0
 
-## Major changes 
+## Major changes
 
-* Internal `Rcpp` functions have been reorganized to remove all dependencies on `dplyr` C++ functions. 
+* Internal `Rcpp` functions have been reorganized to remove all dependencies on `dplyr` C++ functions.
 
 ## Minor changes
 
-* Due to internal refactoring of Rcpp functions, only data.frames containing Numeric, Logical, Integer, Character, and List column types are supported. Columns containing Raw, Complex, or other R classes are not supported and will issue an error. 
+* Due to internal refactoring of Rcpp functions, only data.frames containing Numeric, Logical, Integer, Character, and List column types are supported. Columns containing Raw, Complex, or other R classes are not supported and will issue an error.
 
 * Factors are now disallowed from grouping variables in multiset operations to avoid sort order discrepancies, and compatibility with factor handling in `dplyr` v.0.8.0. Factors will now be internally type-converted to character and a warning is issued.
 
@@ -199,7 +202,7 @@ be suppressed by using `sorted = FALSE`.
 ## Minor changes
 
 * intervals returned from `bed_random()` are sorted by `chrom` and `start` by default.
-  
+
 ## Bug fixes
 
 * Merge intervals in `bed_jaccard()` and use numeric values for calculation (fixes #204).

R/read_bed.r

@@ -174,44 +174,22 @@ bed12_coltypes <- list(
 )
 
 
-#' Import and convert a bigwig file into a valr compatible tbl
+#' Read a bigwig file into a valr compatible tbl
 #'
-#' @description
-#' `r lifecycle::badge("deprecated")`
-#'
-#' This function will output a 5 column tibble with
-#' zero-based chrom, start, end, score, and strand columns.
+#' This function will output a 4 column tibble with
+#' zero-based chrom, start, end, value columns.
 #'
 #' @param path path to bigWig file
-#' @param set_strand strand to add to output (defaults to "+")
-#'
-#' @note This functions uses \code{rtracklayer} to import bigwigs which
-#' has unstable support for the windows platform and therefore may error
-#' for windows users (particularly for 32 bit window users).
+#' @param ... params for `cpp11bigwig::read_bigwig()`
 #'
 #' @examples
-#' \dontrun{
-#'   bw <- read_bigwig(valr_example("hg19.dnase1.bw"))
-#'   head(bw)
-#' }
+#' read_bigwig(valr_example("hg19.dnase1.bw"))
+#'
+#' read_bigwig(valr_example("hg19.dnase1.bw"), as = "GRanges")
+#'
 #' @export
-read_bigwig <- function(path, set_strand = "+") {
-  lifecycle::deprecate_stop(
-    when = "0.8.3",
-    what = "read_bigwig()",
-    details = c(
-      x = paste0(
-        "read_bigwig() was removed because rtracklayer does not pass ",
-        "CRAN AddressSantizer checks of the UCSC C-library code vendored ",
-        "in rtracklayer."
-      ),
-      i = paste0(
-        "use `bigWigToBedGraph` to convert bw to bedGraph,",
-        " and then `read_bedgraph()`. Alternatively use ",
-        "`rtracklayer::import()` then `gr_to_bed()`."
-      )
-    )
-  )
+read_bigwig <- function(path, ...) {
+  cpp11bigwig::read_bigwig(path, ...)
 }
 
 #' Import and convert a GTF/GFF file into a valr compatible bed tbl format

man/read_bigwig.Rd

@@ -45,9 +45,8 @@ test_that("read broadPeak", {
 })
 
 test_that("read bigwig", {
-  skip_on_os("windows")
-  expect_error(read_bigwig(bigwig_path))
-  # expect_equal(ncol(x), 5)
+  x <- read_bigwig(bigwig_path)
+  expect_equal(ncol(x), 4)
 })