fix: macOS bigWigToWig fallback + create_sparse value column (5.11.12)

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: misha
 Title: Toolkit for Analysis of Genomic Data
-Version: 5.11.11
+Version: 5.11.12
 Authors@R: c(
     person("Misha", "Hoichman", , "misha@hoichman.com", role = "aut"),
     person("Aviezer", "Lifshitz", , "aviezer.lifshitz@weizmann.ac.il", role = c("aut", "cre")),

NEWS.md

@@ -1,3 +1,8 @@
+# misha 5.11.12
+
+* `gtrack.import()` no longer fails on non-Linux platforms: the bundled `bigWigToWig` is a Linux x86-64 binary, so macOS/Windows now use one found on `PATH` (or `options(misha.bigWigToWig = "...")`), with an actionable error if none is installed.
+* `gtrack.create_sparse()` can take the values from a `value` column of `intervals` when `values` is omitted, which rules out a values/intervals order mismatch. Documented that `gintervals()` returns its rows sorted, so a separately held value vector must be reordered too.
+
 # misha 5.11.11
 
 * **Internal consistency:** `N`/`*` are now scored the same on both strands in the PWM energy routines (reverse used `log(0.25)`, forward used the column average; both now use the average). No behavior change for genomic scoring, where `N`-windows are masked to `-Inf`; keeps `DnaPSSM` in sync with the `prego` package.

R/intervals-core.R

@@ -105,7 +105,11 @@
 #' @param ends an array of end coordinates. If '-1' chromosome size is assumed.
 #' @param strands 'NULL', a numeric vector of '-1', '0' or '1' values, or a
 #' character/factor vector with values "+", "-", ".", "*" or ""
-#' @return A data frame representing the intervals.
+#' @return A data frame representing the intervals, sorted in the canonical
+#' chromosome order (which is not the order of the arguments). Beware of keeping
+#' per-interval data in a separate parallel vector across a \code{gintervals()}
+#' call: the rows are reordered and the vector is not. Put such data in a column
+#' of the resulting data frame instead.
 #' @seealso \code{\link{gintervals.2d}}, \code{\link{gintervals.force_range}}
 #' @keywords ~intervals
 #' @examples

R/track-create.R

@@ -215,10 +215,20 @@ gtrack.create_pwm_energy <- function(track = NULL, description = NULL, pssmset =
 #' last connected database. Use \code{\link{gdir.cd}} with an absolute path to
 #' change where new tracks are created.
 #'
+#' \code{values} is matched to \code{intervals} row by row, in the order the
+#' intervals are passed; \code{intervals} need not be sorted. Note however that
+#' \code{\link{gintervals}} returns its result sorted in the canonical
+#' chromosome order, so building \code{intervals} with \code{gintervals()} while
+#' keeping \code{values} in the original order will bind values to the wrong
+#' intervals. Keep the values in a \code{value} column of the intervals data
+#' frame and omit \code{values} to make such a mismatch impossible.
+#'
 #' @param track track name
 #' @param description a character string description
 #' @param intervals a set of one-dimensional intervals
-#' @param values an array of numeric values - one for each interval
+#' @param values an array of numeric values - one for each interval, in the same
+#' order as the rows of \code{intervals}. If \code{NULL}, the \code{value}
+#' column of \code{intervals} is used.
 #' @return None.
 #' @seealso \code{\link{gtrack.create}}, \code{\link{gtrack.2d.create}},
 #' \code{\link{gtrack.smooth}}, \code{\link{gtrack.modify}},
@@ -241,13 +251,20 @@ gtrack.create_pwm_energy <- function(track = NULL, description = NULL, pssmset =
 #'
 #' @export gtrack.create_sparse
 gtrack.create_sparse <- function(track = NULL, description = NULL, intervals = NULL, values = NULL) {
-    if (is.null(substitute(track)) || is.null(description) || is.null(intervals) || is.null(values)) {
+    if (is.null(substitute(track)) || is.null(description) || is.null(intervals)) {
         stop("Usage: gtrack.create_sparse(track, description, intervals, values)", call. = FALSE)
     }
     .gcheckroot()
 
     intervals <- rescue_ALLGENOME(intervals, as.character(substitute(intervals)))
 
+    if (is.null(values)) {
+        if (!is.data.frame(intervals) || !("value" %in% colnames(intervals))) {
+            stop("values is missing and intervals has no \"value\" column", call. = FALSE)
+        }
+        values <- intervals$value
+    }
+
     trackstr <- do.call(.gexpr2str, list(substitute(track)), envir = parent.frame())
     .gconfirmtrackcreate(trackstr)
 

R/track-import.R

@@ -1,9 +1,29 @@
 # Track import functions and helpers
 
-get_bigWigToWig_bin <- function() {
-    dir <- tempdir()
-    utils::untar(system.file("bigWigToWig.tar.gz", package = "misha"), exdir = dir)
-    return(file.path(dir, "bigWigToWig"))
+get_bigWigToWig_bin <- function(sysname = Sys.info()[["sysname"]]) {
+    bin <- getOption("misha.bigWigToWig", "")
+    if (nzchar(bin)) {
+        return(bin)
+    }
+
+    # ponytail: only a Linux x86-64 binary is bundled; elsewhere fall back to PATH
+    if (sysname == "Linux") {
+        dir <- tempdir()
+        utils::untar(system.file("bigWigToWig.tar.gz", package = "misha"), exdir = dir)
+        return(file.path(dir, "bigWigToWig"))
+    }
+
+    bin <- Sys.which("bigWigToWig")
+    if (!nzchar(bin)) {
+        stop(sprintf(paste0(
+            "The bigWigToWig executable bundled with misha is a Linux x86-64 binary and cannot run on %s.\n",
+            "Install UCSC's bigWigToWig for your platform, e.g.:\n",
+            "  conda install -c bioconda ucsc-bigwigtowig\n",
+            "  or download it from http://hgdownload.soe.ucsc.edu/admin/exe/\n",
+            "Then put it on PATH, or set options(misha.bigWigToWig = \"/path/to/bigWigToWig\")."
+        ), sysname), call. = FALSE)
+    }
+    bin
 }
 
 

tests/testthat/test-bigwigtowig-resolution.R

@@ -0,0 +1,36 @@
+# The bundled bigWigToWig is a Linux x86-64 binary, so bigWig import cannot work
+# on other platforms without an externally installed converter. There is no
+# bigWig import test, which is why this went unnoticed on the macOS CI runner.
+
+test_that("options(misha.bigWigToWig) wins over everything else", {
+    withr::local_options(misha.bigWigToWig = "/some/where/bigWigToWig")
+    expect_equal(get_bigWigToWig_bin(), "/some/where/bigWigToWig")
+    # honored on every platform, including the bundled-binary one
+    expect_equal(get_bigWigToWig_bin("Linux"), "/some/where/bigWigToWig")
+})
+
+test_that("the bundled binary is used on Linux", {
+    skip_if_not(Sys.info()[["sysname"]] == "Linux")
+    withr::local_options(misha.bigWigToWig = NULL)
+    expect_true(file.exists(get_bigWigToWig_bin("Linux")))
+})
+
+test_that("a converter on PATH is used off-Linux", {
+    withr::local_options(misha.bigWigToWig = NULL)
+    dir <- withr::local_tempdir()
+    fake <- file.path(dir, "bigWigToWig")
+    writeLines("#!/bin/sh\nexit 0", fake)
+    Sys.chmod(fake, "0755")
+    withr::local_envvar(PATH = dir)
+
+    expect_equal(normalizePath(get_bigWigToWig_bin("Darwin")), normalizePath(fake))
+})
+
+test_that("off-Linux with no converter errors with install instructions", {
+    withr::local_options(misha.bigWigToWig = NULL)
+    withr::local_envvar(PATH = withr::local_tempdir()) # empty -> Sys.which finds nothing
+
+    expect_error(get_bigWigToWig_bin("Darwin"), "cannot run on Darwin")
+    expect_error(get_bigWigToWig_bin("Darwin"), "ucsc-bigwigtowig")
+    expect_error(get_bigWigToWig_bin("Darwin"), "misha\\.bigWigToWig")
+})

tests/testthat/test-gtrack.create_sparse-value-alignment.R

@@ -0,0 +1,54 @@
+create_isolated_test_db()
+
+# Reported by a user who built intervals with gintervals() (which sorts) while
+# keeping values in the pre-sort order. create_sparse itself binds values by the
+# original row index, so unsorted input is fine.
+test_that("create_sparse binds values by original row order, not sorted order", {
+    tmptrack <- paste0("test.tmptrack_", sample(1:1e9, 1))
+    gtrack.rm(tmptrack, force = TRUE)
+    withr::defer(gtrack.rm(tmptrack, force = TRUE))
+
+    # deliberately out of canonical chrom order
+    intervals <- data.frame(
+        chrom = c("chr2", "chr1", "chr2", "chr1"),
+        start = c(100, 200, 300, 400),
+        end = c(101, 201, 301, 401),
+        stringsAsFactors = FALSE
+    )
+    values <- c(2.1, 1.1, 2.3, 1.4)
+
+    gtrack.create_sparse(tmptrack, "Test", intervals, values)
+
+    r <- gextract(tmptrack, gintervals.all(), colnames = "value")
+    expected <- intervals[order(intervals$chrom, intervals$start), ]
+    expect_equal(as.character(r$chrom), expected$chrom)
+    expect_equal(r$start, expected$start)
+    expect_equal(r$value, values[order(intervals$chrom, intervals$start)], tolerance = 1e-6)
+})
+
+test_that("create_sparse takes values from the value column when values is omitted", {
+    tmptrack <- paste0("test.tmptrack_", sample(1:1e9, 1))
+    gtrack.rm(tmptrack, force = TRUE)
+    withr::defer(gtrack.rm(tmptrack, force = TRUE))
+
+    intervals <- data.frame(
+        chrom = c("chr2", "chr1", "chr2"),
+        start = c(100, 200, 300),
+        end = c(101, 201, 301),
+        value = c(2.1, 1.1, 2.3),
+        stringsAsFactors = FALSE
+    )
+
+    gtrack.create_sparse(tmptrack, "Test", intervals)
+
+    r <- gextract(tmptrack, gintervals.all(), colnames = "value")
+    expect_equal(r$value, c(1.1, 2.1, 2.3), tolerance = 1e-6)
+})
+
+test_that("create_sparse errors when values is omitted and there is no value column", {
+    intervals <- gintervals(1, 100, 200)
+    expect_error(
+        gtrack.create_sparse("test.no_value_col", "Test", intervals),
+        "no \"value\" column"
+    )
+})

tests/testthat/test-gvtrack-clear.R

@@ -1,3 +1,5 @@
+create_isolated_test_db()
+
 test_that("gvtrack.clear removes all virtual tracks for the current working directory", {
     gvtrack.create("vt_clear_1", "test.fixedbin", "max")
     gvtrack.create("vt_clear_2", "test.fixedbin", "avg")