1.2.8.2

Minor bug fixes and updates with precursor mass calibrations.
qzhang503 · Jun 27, 2023 · 10651ee · 10651ee
1 parent b487e01
commit 10651ee
Show file tree

Hide file tree

Showing 35 changed files with 1,004 additions and 392 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: mzion
 Type: Package
-Title: Database Searches of Proteomic Mass-spectrometrirc Data
-Version: 1.2.8.1
+Title: Proteomics Database Searches of Mass-spectrometrirc Data
+Version: 1.2.8.2
 Authors@R: 
     person(given = "Qiang",
            family = "Zhang",

diff --git a/R/mgfs.R b/R/mgfs.R
@@ -1598,13 +1598,49 @@ prepBrukerMGF <- function (file = NULL, begin_offset = 5L, charge_offset = 5L)
   message("Processing: ", file)
 
   lines <- readLines(file)
+  hd <- lines[1:100]
 
+  ### Some MGFs may have additional "SCANS=" lines
+  pat_sc <- "SCANS"
+  sc <- .Internal(which(stringi::stri_startswith_fixed(hd, pat_sc)))
+
+  if (length(sc)) {
+    lscs  <- .Internal(which(stringi::stri_startswith_fixed(lines, pat_sc)))
+    lines <- lines[-lscs]
+    rm(list = "lscs")
+  }
+  rm(list = c("pat_sc", "sc"))
+
+  ###PrecursorID: 
+  pat_pr <- "###PrecursorID"
+  pr <- .Internal(which(stringi::stri_startswith_fixed(hd, pat_pr)))
+
+  if (length(pr)) {
+    lprs  <- .Internal(which(stringi::stri_startswith_fixed(lines, pat_pr)))
+    lines <- lines[-lprs]
+    rm(list = "lprs")
+  }
+  rm(list = c("pat_pr", "pr"))
+
+  ###CCS: 
+  pat_ccs <- "###CCS"
+  ccs <- .Internal(which(stringi::stri_startswith_fixed(hd, pat_ccs)))
+
+  if (length(ccs)) {
+    ccs  <- .Internal(which(stringi::stri_startswith_fixed(lines, pat_ccs)))
+    lines <- lines[-ccs]
+    rm(list = "ccs")
+  }
+  rm(list = c("pat_ccs", "ccs"))
+
+  ## Processing
   begins <- .Internal(which(stringi::stri_startswith_fixed(lines, "BEGIN IONS")))
   ends   <- .Internal(which(stringi::stri_endswith_fixed(lines, "END IONS")))
   hdrs   <- 1:(begins[1]-begin_offset-1L)
 
   zls <- lines[begins+5L]
-  oks <- grepl("^CHARGE", zls) & (zls != "CHARGE=1+")
+  # oks <- grepl("^CHARGE", zls) & (zls != "CHARGE=1+")
+  oks <- grepl("^CHARGE", zls) & (zls != "CHARGE=1+") & grepl("^###Mobility", lines[begins-1L])
   rm(list = "zls")
 
   b_oks <- begins[oks] - begin_offset
@@ -1623,7 +1659,7 @@ prepBrukerMGF <- function (file = NULL, begin_offset = 5L, charge_offset = 5L)
 #' @param filepath A file path to MGF.
 #' @param n_cores The number of CPU cores.
 #' @export
-mprepBrukerMGF <- function (filepath, n_cores = 48L) 
+mprepBrukerMGF <- function (filepath, n_cores = 32L) 
 {
   message("Preparing Bruker's MGFs.")
 

diff --git a/R/ms2_gen.R b/R/ms2_gen.R
@@ -100,8 +100,6 @@ gen_ms2ions_base <- function (aa_seq = NULL, ms1_mass = NULL,
                               mod_indexes = NULL, 
                               type_ms2ions = "by", maxn_vmods_per_pep = 5L, 
                               maxn_sites_per_vmod = 3L, 
-
-                              # dummy
                               maxn_fnl_per_seq = 3L, maxn_vnl_per_seq = 3L, 
                               maxn_vmods_sitescombi_per_pep = 64L) 
 {
@@ -182,11 +180,13 @@ gen_ms2ions_a0_vnl0_fnl1 <- function (aa_seq, ms1_mass = NULL,
                                       maxn_vmods_per_pep = 5L, 
                                       maxn_sites_per_vmod = 3L, 
                                       maxn_fnl_per_seq = 3L, 
-
-                                      # dummy
                                       maxn_vnl_per_seq = 3L, 
                                       maxn_vmods_sitescombi_per_pep = 64L) 
 {
+  # cannot distinguish a neutral loss is in-source or in-MS2: 
+  # "+ms1_offset and +ms2_nl" versus "-ms1_offset and -ms2_nl" 
+  # if (ms1_offset != 0) maxn_vnl_per_seq = 1L
+
   if (maxn_fnl_per_seq < 2L)
     return(
       gen_ms2ions_base(aa_seq = aa_seq, ms1_mass = ms1_mass, 
@@ -478,10 +478,8 @@ gen_ms2ions_a1_vnl0_fnl0 <- function (aa_seq, ms1_mass = NULL, aa_masses = NULL,
                                       mod_indexes = NULL, type_ms2ions = "by", 
                                       maxn_vmods_per_pep = 5L, 
                                       maxn_sites_per_vmod = 3L, 
-
-                                      # dummy
-                                      maxn_fnl_per_seq = 3L, maxn_vnl_per_seq = 3L, 
-
+                                      maxn_fnl_per_seq = 3L, 
+                                      maxn_vnl_per_seq = 3L, 
                                       maxn_vmods_sitescombi_per_pep = 64L) 
 {
   aas <- .Internal(strsplit(aa_seq, "", fixed = TRUE, perl = FALSE, useBytes = FALSE))
@@ -770,8 +768,6 @@ gen_ms2ions_a1_vnl0_fnl1 <- function (aa_seq = NULL, ms1_mass = NULL,
                                       maxn_sites_per_vmod = 3L, 
                                       maxn_vmods_sitescombi_per_pep = 64L, 
                                       maxn_fnl_per_seq = 3L, 
-
-                                      # dummy
                                       maxn_vnl_per_seq = 3L) 
 {
   if (maxn_fnl_per_seq < 2L)
@@ -1146,8 +1142,6 @@ gen_ms2ions_a1_vnl1_fnl0 <- function (aa_seq = NULL, ms1_mass = NULL,
                                       maxn_vmods_per_pep = 5L, 
                                       maxn_sites_per_vmod = 3L, 
                                       maxn_vmods_sitescombi_per_pep = 64L, 
-
-                                      # dummy
                                       maxn_fnl_per_seq = 3L, 
                                       maxn_vnl_per_seq = 3L) 
 {