Merge pull request #341 from rformassspectrometry/jomain

Support chunk-wise processing in containsMz

.editorconfig

@@ -6,7 +6,7 @@ root = true
 charset = utf-8
 end_of_line = lf
 trim_trailing_whitespace = true
-insert_final_newline = false
+insert_final_newline = true
 
 [*.R]
 indent_style = space
@@ -22,4 +22,4 @@ indent_style = tab
 
 [*.yml]
 indent_style = space
-indent_size = 2
+indent_size = 2

.github/workflows/check-bioc.yml

@@ -53,7 +53,7 @@ jobs:
       fail-fast: false
       matrix:
         config:
-          - { os: ubuntu-latest, r: '4.4', bioc: '3.20', cont: "bioconductor/bioconductor_docker:devel", rspm: "https://packagemanager.rstudio.com/cran/__linux__/focal/latest" }
+          - { os: ubuntu-latest, r: 'devel', bioc: 'devel', cont: "bioconductor/bioconductor_docker:devel", rspm: "https://packagemanager.rstudio.com/cran/__linux__/focal/latest" }
           - { os: macOS-latest, r: '4.4', bioc: '3.20'}
           - { os: windows-latest, r: '4.4', bioc: '3.20'}
     env:
@@ -239,7 +239,7 @@ jobs:
       - name: Test coverage
         if: github.ref == 'refs/heads/main' && env.run_covr == 'true' && runner.os == 'Linux'
         run: |
-          covr::codecov()
+          covr::codecov(token = "${{ secrets.CODECOV_TOKEN }}")
         shell: Rscript {0}
 
       - name: Install package

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: Spectra
 Title: Spectra Infrastructure for Mass Spectrometry Data
-Version: 1.15.13
+Version: 1.17.1
 Description: The Spectra package defines an efficient infrastructure
    for storing and handling mass spectrometry spectra and functionality to
    subset, process, visualize and compare spectra data. It provides different

NEWS.md

@@ -1,3 +1,9 @@
+# Spectra 1.17
+
+## Changes in 1.17.1
+
+- Refactor `containsMz()` to support chunk-wise processing.
+
 # Spectra 1.15
 
 ## Changes in 1.15.13

R/Spectra-functions.R

@@ -63,7 +63,13 @@ NULL
 #' @description
 #'
 #' This function applies the processing queue and an arbitrary function to
-#' the peaks matrix of each spectrum of the `Spectra` object `object`.
+#' the peaks matrix of each spectrum of the `Spectra` object `object`. It has
+#' build-in parallel and/or chunk-wise processing enabled through parameter
+#' `f`, that allows to define how the `Spectra` (or rather its backend) needs
+#' to be splitted. The default `f = .parallel_processing_factor(object)` splits
+#' the backend by chunk (if a finite chunk size is defined for the `Spectra`)
+#' or by it's optimal parallel processing factor. See the description of
+#' the `.parallel_processing_factor()` function below for information.
 #'
 #' @param object `Spectra` object.
 #'
@@ -78,7 +84,8 @@ NULL
 #'
 #' @param f `factor` or `vector` that can be coerced to one defining how the
 #'     data should be split for parallel processing. Set to `NULL` or
-#'     `factor()` to disable splitting and parallel processing.
+#'     `factor()` to disable splitting and parallel processing. See function
+#'     description above for details and information.
 #'
 #' @param columns `character` defining the columns that should be returned.
 #'     This will be passed to the backend's `peaksData` function.
@@ -571,39 +578,8 @@ combineSpectra <- function(x, f = x$dataStorage, p = x$dataStorage,
 
 #' @description
 #'
-#' Internal function to check if any (or all) of the provided `mz` values are
-#' in the spectras' m/z.
-#'
-#' @param x `Spectra` object
-#'
-#' @param mz `numeric` of m/z value(s) to check in each spectrum of `x`.
-#'
-#' @param tolarance `numeric(1)` with the tolerance.
-#'
-#' @param ppm `numeric(1)` with the ppm.
-#'
-#' @param condFun `function` such as `any` or `all`.
-#'
-#' @param parallel `BiocParallel` parameter object.
-#'
-#' @return `logical` same length than `x`.
-#'
-#' @author Johannes Rainer
-#'
-#' @importFrom MsCoreUtils common
-#'
-#' @noRd
-.has_mz <- function(x, mz = numeric(), tolerance = 0, ppm = 20, condFun = any,
-                    parallel = SerialParam()) {
-    mzs <- mz(x, BPPARAM = parallel)
-    vapply(mzs, FUN = function(z)
-        condFun(common(mz, z, tolerance = tolerance, ppm = ppm)), logical(1))
-}
-
-#' @description
-#'
-#' Same as `.has_mz` only that a different `mz` is used for each spectrum in
-#' `x`. Length of `mz` is thus expected to be equal to length of `x`.
+#' Check for presence of an m/z value in each spectrum. Each spectrum gets
+#' its own m/z.
 #'
 #' @param mz `numeric` **same length as `x`**.
 #'

R/Spectra.R

@@ -3278,23 +3278,18 @@ setMethod("containsMz", "Spectra", function(object, mz = numeric(),
                                             tolerance = 0,
                                             ppm = 20, which = c("any", "all"),
                                             BPPARAM = bpparam()) {
-    cond_fun <- match.fun(match.arg(which))
-    if (all(is.na(mz)))
-        return(rep(NA, length(object)))
-    mz <- unique(sort(mz))
-    BPPARAM <- backendBpparam(object@backend, BPPARAM)
-    ## TODO: fix to use .peaksapply instead.
-    if (is(BPPARAM, "SerialParam"))
-        .has_mz(object, mz, tolerance = tolerance, ppm = ppm,
-                condFun = cond_fun, parallel = BPPARAM)
-    else {
-        sp <- SerialParam()
-        f <- as.factor(dataStorage(object))
-        res <- .lapply(object, FUN = .has_mz, mz = mz, tolerance = tolerance,
-                       condFun = cond_fun, parallel = sp, f = f,
-                       BPPARAM = BPPARAM)
-        unsplit(res, f = f)
-    }
+    if (length(object)) {
+        cond_fun <- match.fun(match.arg(which))
+        if (all(is.na(mz)))
+            return(rep(NA, length(object)))
+        mz <- unique(sort(mz))
+        BPPARAM <- backendBpparam(object@backend, BPPARAM)
+        unlist(.peaksapply(
+            object, FUN = .peaks_contain_mz, mz = mz, tolerance = tolerance,
+            ppm = ppm, condFun = cond_fun, BPPARAM = BPPARAM),
+            use.names = FALSE
+            )
+    } else logical()
 })
 
 #' @rdname addProcessing
@@ -3327,12 +3322,12 @@ setMethod("containsNeutralLoss", "Spectra", function(object, neutralLoss = 0,
 #' @export
 setMethod("entropy", "Spectra", function(object, normalized = TRUE) {
     if (length(object)) {
-    if (normalized) entropy_fun <- nentropy
-    else entropy_fun <- entropy
-    unlist(.peaksapply(
-        object, FUN = function(pks, ...) entropy_fun(pks[, "intensity"])),
-        use.names = FALSE
-        )
+        if (normalized) entropy_fun <- nentropy
+        else entropy_fun <- entropy
+        unlist(.peaksapply(
+            object, FUN = function(pks, ...) entropy_fun(pks[, "intensity"])),
+            use.names = FALSE
+            )
     } else numeric()
 })
 #' @rdname addProcessing

R/peaks-functions.R

@@ -737,3 +737,13 @@ joinPeaksNone <- function(x, y, ...) {
     if (keep) x[sel, , drop = FALSE]
     else x[!sel, , drop = FALSE]
 }
+
+#' Check for presence of peaks defined by their m/z value. Note that this
+#' function does **not** return a peak matrix, but only a logical of length 1!
+#'
+#' @return `logical(1)`
+#' @noRd
+.peaks_contain_mz <- function(x, mz = numeric(), tolerance = 0, ppm = 20,
+                              condFun = any, ...) {
+    condFun(common(mz, x[, "mz"], tolerance = tolerance, ppm = ppm))
+}