Merge pull request #305 from rformassspectrometry/phili

fix: `bin` function

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: Spectra
 Title: Spectra Infrastructure for Mass Spectrometry Data
-Version: 1.11.11
+Version: 1.11.12
 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

@@ -1,5 +1,11 @@
 # Spectra 1.11
 
+## Changes in 1.11.12
+
+- Fix issue with `bin` function (see 
+  [issue #302](https://github.com/rformassspectrometry/Spectra/issues/303)). 
+  Addition of `zero.rm` parameter to prevent creation of *empty* bins. 
+
 ## Changes in 1.11.11
 
 - Fix issue with `filterFourierTransformArtefacts` function (see [issue

R/Spectra.R

@@ -1049,6 +1049,10 @@ NULL
 #' @param z For `filterPrecursorCharge`: `integer()` with the precursor charges
 #'     to be used as filter.
 #'
+#' @param zero.rm `logical`. For `bin`: indicating whether to remove bins with
+#'     zero intensity. Defaults to `TRUE`, meaning the function will discard
+#'     bins created with an intensity of 0 to enhance memory efficiency.
+#'
 #' @param ... Additional arguments.
 #'
 #' @author Sebastian Gibb, Johannes Rainer, Laurent Gatto
@@ -2437,7 +2441,7 @@ setMethod("reset", "Spectra", function(object, ...) {
 #' @exportMethod bin
 setMethod("bin", "Spectra", function(x, binSize = 1L, breaks = NULL,
                                      msLevel. = uniqueMsLevels(x),
-                                     FUN = sum) {
+                                     FUN = sum, zero.rm = TRUE) {
     if (!.check_ms_level(x, msLevel.))
         return(x)
     if (!length(breaks)) {
@@ -2449,7 +2453,7 @@ setMethod("bin", "Spectra", function(x, binSize = 1L, breaks = NULL,
     }
     mids <- (breaks[-length(breaks)] + breaks[-1L]) / 2
     x <- addProcessing(x, .peaks_bin, breaks = breaks, mids = mids,
-                       agg_fun = FUN, msLevel = msLevel.,
+                       agg_fun = FUN, msLevel = msLevel., zero.rm = zero.rm,
                        spectraVariables = "msLevel")
     x@processing <- .logging(x@processing,
                              "Spectra of MS level(s) ",

R/peaks-functions.R

@@ -180,6 +180,11 @@ NULL
 #'
 #' @param mids mid points. This parameter is mandatory.
 #'
+#' @param zero.rm `logical`  indicating whether to remove bins with zero
+#'     intensity. Defaults to `TRUE`, meaning the function will discard bins
+#'     created with an intensity of 0 to enhance memory efficiency.
+#'
+#'
 #' @inheritParams .peaks_remove
 #'
 #' @return `matrix` with columns `"mz"` and `"intensity"`
@@ -191,12 +196,15 @@ NULL
                                     by = binSize),
                        agg_fun = sum,
                        mids,
-                       msLevel = spectrumMsLevel, ...) {
+                       msLevel = spectrumMsLevel, zero.rm = TRUE, ...) {
     if (!(spectrumMsLevel %in% msLevel))
         return(x)
     bins <- MsCoreUtils::bin(x[, 2], x[, 1], size = binSize, breaks = breaks,
                              FUN = agg_fun, returnMids = FALSE)
-    cbind(mz = mids, intensity = bins)
+    if (zero.rm) {
+        keep <- which(bins != 0)
+        cbind(mz = mids[keep], intensity = bins[keep])
+    } else cbind(mz = mids, intensity = bins)
 }
 
 #' @importFrom stats quantile

tests/testthat/test_Spectra.R

@@ -1105,17 +1105,17 @@ test_that("filterRt,Spectra works", {
 test_that("bin,Spectra works", {
     sps <- Spectra(tmt_mzr)
     pks <- peaksData(sps)
-    res <- bin(sps, binSize = 2)
+    res <- bin(sps, binSize = 2, zero.rm = FALSE)
     expect_true(length(res@processingQueue) == 1)
-    res1 <- bin(sps, msLevel = 1, binSize = 2)
+    res1 <- bin(sps, msLevel = 1, binSize = 2, zero.rm = FALSE)
 
     expect_identical(peaksData(res1)[res1$msLevel == 2],
                      pks[sps$msLevel == 2])
 
     mzr <- range(unlist(mz(sps)))
     brks <- MsCoreUtils:::.fix_breaks(
                               seq(floor(mzr[1]), ceiling(mzr[2]), by = 2), mzr)
-    res1 <- bin(sps, msLevel = 1, breaks = brks)
+    res1 <- bin(sps, msLevel = 1, breaks = brks, zero.rm = FALSE)
     res1_pks <- peaksData(res1)
     res_pks <- peaksData(res)
     expect_identical(res1_pks[res1$msLevel == 1],

tests/testthat/test_peaks-functions.R

@@ -75,30 +75,43 @@ test_that(".peaks_bin works", {
     brks <- seq(min(x), max(x), by = 1L)
     brks <- MsCoreUtils:::.fix_breaks(brks, range(x))
     mids <- seq_len(length(brks) -1)
-    res <- .peaks_bin(x, spectrumMsLevel = 1L, breaks = brks, mids = mids)
+    res <- .peaks_bin(x, spectrumMsLevel = 1L, breaks = brks, mids = mids,
+                      zero.rm = FALSE)
     expect_identical(res[-1, 2], x[, 2])
     expect_identical(res[, 1], as.numeric(mids))
 
     brks <- seq(min(x), max(x), by = 2L)
     brks <- MsCoreUtils:::.fix_breaks(brks, range(x))
     mids <- (brks[-length(brks)] + brks[-1L]) / 2
     res <- .peaks_bin(x, spectrumMsLevel = 1L, breaks = brks,
-                                mids = mids)
+                                mids = mids, zero.rm = FALSE)
     expect_equal(res[, 1], seq(1, 25, by = 2))
     expect_equal(res[, 2], c(0, 3, 4, 0, 0, 4, 16, 3, 1, 2, 1, 15, 6))
     res <- .peaks_bin(x, spectrumMsLevel = 1L, breaks = brks,
-                      agg_fun = max, mids = mids)
+                      agg_fun = max, mids = mids, zero.rm = FALSE)
     expect_equal(res[, 1], seq(1, 25, by = 2))
     expect_equal(res[, 2], c(0, 2, 3, 0, 0, 3, 10, 2, 1, 2, 1, 10, 5))
-    res <- .peaks_bin(x, spectrumMsLevel = 1L, msLevel = 2L, binSize = 2L)
+    res <- .peaks_bin(x, spectrumMsLevel = 1L, msLevel = 2L, binSize = 2L,
+                      zero.rm = FALSE)
     expect_identical(res, x)
 
     brks <- seq(0.5, 30.5, by = 2)
     mids <- seq((length(brks) - 1))
     res <- Spectra:::.peaks_bin(x, breaks = brks, mids = mids,
-                                spectrumMsLevel = 1L)
+                                spectrumMsLevel = 1L, zero.rm = FALSE)
     expect_equal(res[, 1], mids)
     expect_equal(res[, 2], c(1, 5, 1, 0, 1, 13, 8, 1, 3, 0, 6, 15, 1, 0, 0))
+
+    breaks <- seq(0, 1000, by = 0.01)
+    mids <- (breaks[-length(breaks)] + breaks[-1L]) / 2
+    res_true <- Spectra:::.peaks_bin(sciex_pks[[1]], breaks = breaks, mids = mids,
+                                spectrumMsLevel = 1L, zero.rm =  TRUE)
+    res_false <- Spectra:::.peaks_bin(sciex_pks[[1]], breaks = breaks, mids = mids,
+                                      spectrumMsLevel = 1L, zero.rm =  FALSE)
+    expect_true(all(res_true[,"intensity"] != 0))
+    expect_false(all(res_false[,"intensity"] != 0))
+    expect_false(length(res_true) == length(res_false))
+
 })
 
 test_that("joinPeaksNone works", {

vignettes/Spectra.Rmd

@@ -955,10 +955,11 @@ Another way of comparing spectra would be to *bin* the spectra and to cluster
 them based on similar intensity values. Spectra binning ensures that the binned
 m/z values are comparable across all spectra. Below we bin our spectra using a
 bin size of 0.1 (i.e. all peaks with an m/z smaller than 0.1 are aggregated into
-one binned peak.
+one binned peak. Below, we explicitly set `zero.rm = FALSE` to retain all bins
+generated by the function, including those with an intensity of zero. 
 
 ```{r}
-sps_bin <- Spectra::bin(sps, binSize = 0.1)
+sps_bin <- Spectra::bin(sps, binSize = 0.1, zero.rm = FALSE)
 ```
 
 All spectra will now have the same number of m/z values.