From 676c56e008626532b84050470c4640114260752d Mon Sep 17 00:00:00 2001
From: qzhang503 <qzhang503@gmail.com>
Date: Sun, 3 Dec 2023 07:57:27 -0600
Subject: [PATCH] 2023-12-03

---
 R/ms2frames.R       |  8 +++++---
 R/msmsmatches2.R    |  5 +++--
 R/utils_os.R        | 38 +++++++++++++++++++-------------------
 man/delete_files.Rd |  4 +++-
 4 files changed, 30 insertions(+), 25 deletions(-)

diff --git a/R/ms2frames.R b/R/ms2frames.R
index c9105f3..cff3fec 100644
--- a/R/ms2frames.R
+++ b/R/ms2frames.R
@@ -96,6 +96,7 @@ hpair_mgths <- function (ms1_offset = 0, notch = NULL, mgfs, n_modules,
   else {
     mgfs <- make_dia_mgfs(mgfs = mgfs, mgf_path = mgf_path, quant = quant, 
                           min_mass = min_mass, ppm_ms1_bin = ppm_ms1_bin)
+    mgfs <- mgfs[names(mgfs) != "-Inf"] # unknown precursors
   }
 
   # to chunks: each chunk has multiple frames: each frame multiple precursors
@@ -265,13 +266,14 @@ make_dia_mgfs <- function (mgfs, mgf_path, quant = "none", min_mass = 200L,
   data_ms2 <- data_ms2[seqs, ]
   mgfs <- dplyr::bind_cols(datalist, dataflat, data_ms2)
 
-  # rm(list = c("datalist", "dataflat", "scans", "data_ms2", 
-  #             "col_nms", "cols_list", "cols_flat", "cols_ms2"))
-  
   ord <- order(mgfs$ms1_mass)
   mgfs <- mgfs[ord, ]
   ms1_bins <- ceiling(log(mgfs$ms1_mass/min_mass)/log(1+ppm_ms1_bin/1e6))
   
+  # ms1_bins can be -Inf or NA since
+  # (1) ms1_moverz can be NA because not yet determined by the current algorithm
+  # (2) ms1_mass can also be NA due to ms1_charge being NA
+  # (3) ms1_charge can be zero since undetermined
   mgfs <- split(mgfs, ms1_bins)
 }
 
diff --git a/R/msmsmatches2.R b/R/msmsmatches2.R
index 0e88976..d5d655c 100644
--- a/R/msmsmatches2.R
+++ b/R/msmsmatches2.R
@@ -98,12 +98,13 @@ ms2match <- function (mgf_path, aa_masses_all, out_path, .path_bin,
   
   delete_files(
     out_path, 
-    ignores = c(mgf_path, "\\.[Rr]$", "\\.(mgf|MGF)$", "\\.(mzML|mzml)$", 
+    ignores = c("\\.[Rr]$", "\\.(mgf|MGF)$", "\\.(mzML|mzml)$", 
                 "\\.xlsx$", "\\.xls$", "\\.csv$", "\\.txt$", "\\.pars$", 
                 "^mgf$", "^mgfs$", "^mzML$", "^mzMLs$", 
                 "Calls", "^PSM$", "^Peptide$", "^Protein$", 
                 "fraction_scheme.rda", "label_scheme.rda", 
-                "label_scheme_full.rda"))
+                "label_scheme_full.rda"), 
+    paths_excluded = mgf_path)
 
   # pairs expts and theos
   files_a  <-  list.files(mgf_path, pattern = "^expttheo_", full.names = TRUE)
diff --git a/R/utils_os.R b/R/utils_os.R
index 54814e6..0b18621 100644
--- a/R/utils_os.R
+++ b/R/utils_os.R
@@ -557,8 +557,9 @@ match_calltime <- function (path = "~/mzion/.MSearches/Cache/Calls",
 #' 
 #' @param path A file path.
 #' @param ignores The file extensions to be ignored.
+#' @param paths_excluded The path to be ignored.
 #' @param ... Arguments for file.remove.
-delete_files <- function (path, ignores = NULL, ...) 
+delete_files <- function (path, ignores = NULL, paths_excluded = NULL, ...) 
 {
   
   # Hadley Ch.19 Quasiquotation, note 97
@@ -577,27 +578,26 @@ delete_files <- function (path, ignores = NULL, ...)
   
   nms <- do.call(list.files, args)
   
-  if (!is.null(ignores)) {
-    nms <- local({
-      dirs <- list.dirs(path, full.names = FALSE, recursive = recursive)
-      dirs <- dirs[dirs != ""]
+  if (length(nms) && length(paths_excluded)) {
+    ignore_case <- if (Sys.info()['sysname'] == "Windows") TRUE else FALSE
+    
+    for (i in seq_along(paths_excluded))
+      nms <- nms[!grepl(paths_excluded[[i]], nms, ignore.case = ignore_case)]
+  }
 
-      idxes_kept <- purrr::map_lgl(dirs, function (x) any(grepl(x, ignores)))
-      
-      nms_kept <- list.files(path = file.path(path, dirs[idxes_kept]),
-                             recursive = TRUE, full.names = TRUE)
-      
-      nms[! nms %in% nms_kept]
-    })
+  if (!is.null(ignores)) {
+    dirs <- list.dirs(path, full.names = FALSE, recursive = recursive)
+    dirs <- dirs[dirs != ""]
+    oks <- purrr::map_lgl(dirs, function (x) any(grepl(x, ignores)))
+    nms_oks <- list.files(path = file.path(path, dirs[oks]),recursive = TRUE, 
+                          full.names = TRUE)
+    nms <- nms[!nms %in% nms_oks]
     
-    nms <- local({
-      exts <- gsub("^.*(\\.[^.]*)$", "\\1", nms)
-      idxes_kept <- purrr::map_lgl(exts, function (x) any(grepl(x, ignores)))
-      
-      nms[!idxes_kept]
-    })
+    exts <- gsub("^.*(\\.[^.]*)$", "\\1", nms)
+    oks <- purrr::map_lgl(exts, function (x) any(grepl(x, ignores)))
+    nms <- nms[!oks]
   }
-  
+
   if (length(nms)) 
     suppressMessages(file.remove(file.path(nms)))
 
diff --git a/man/delete_files.Rd b/man/delete_files.Rd
index 8bcb081..18ff54e 100644
--- a/man/delete_files.Rd
+++ b/man/delete_files.Rd
@@ -4,13 +4,15 @@
 \alias{delete_files}
 \title{Deletes files under a directory.}
 \usage{
-delete_files(path, ignores = NULL, ...)
+delete_files(path, ignores = NULL, paths_excluded = NULL, ...)
 }
 \arguments{
 \item{path}{A file path.}
 
 \item{ignores}{The file extensions to be ignored.}
 
+\item{paths_excluded}{The path to be ignored.}
+
 \item{...}{Arguments for file.remove.}
 }
 \description{