2023-06-04

Codes optimization in matching experimental and theoretical features.
qzhang503 · Jun 4, 2023 · af4608e · af4608e
1 parent b2bcd7d
commit af4608e
Show file tree

Hide file tree

Showing 14 changed files with 55 additions and 103 deletions.
diff --git a/R/ion_ladder.R b/R/ion_ladder.R
@@ -159,8 +159,7 @@ bions_base <- function (aam, tmass, digits = 4L)
   ions <- c(tmass, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -174,8 +173,7 @@ yions_base <- function (aam, tmass, digits = 4L)
   ions <- c(tmass, aam[length(aam):1L])
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -196,8 +194,7 @@ bstarions <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 17.026549, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -219,8 +216,7 @@ b0ions <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 18.010565, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -248,8 +244,7 @@ ystarions <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 17.026549, aam[length(aam):1L])
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -269,8 +264,7 @@ y0ions <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 18.010565, aam[length(aam):1L])
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -290,8 +284,7 @@ cions_base <- function (aam, tmass, digits = 4L)
   ions <- c(tmass + 17.026549, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -311,8 +304,7 @@ zions_base <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 17.026549, aam[length(aam):1L])
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -332,8 +324,7 @@ aions_base <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 27.9949146, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -354,8 +345,7 @@ astarions <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 45.0214636, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -376,8 +366,7 @@ a0ions <- function (aam, tmass, digits = 4L)
   ions <- c(tmass - 46.0054796, aam)
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 
@@ -398,8 +387,7 @@ xions_base <- function (aam, tmass, digits = 4L)
   ions <- c(tmass + 25.9792646, aam[length(aam):1L])
   ions <- cumsum(ions)
   ions <- ions[-1]
-
-  round(ions, digits = digits)
+  # round(ions, digits = digits)
 }
 
 

diff --git a/R/ms2frames.R b/R/ms2frames.R
@@ -1039,7 +1039,6 @@ find_ms2_bypep <- function (theos = NULL, expts = NULL, ex = NULL, d = NULL,
                                ppm_ms2 = 10L, min_ms2mass = 115L, minn_ms2 = 6L, 
                                index_mgf_ms2 = FALSE) 
 {
-  ##############################################################################
   # `theos`
   #   the same pep_seq at different applicable ivmods and NLs
   # 
@@ -1067,8 +1066,7 @@ find_ms2_bypep <- function (theos = NULL, expts = NULL, ex = NULL, d = NULL,
   # (i) %in% and %fin% only shows the first match for duplicated entries thi:
   #   match(1:4, c(1, 2, 2, 5))
   #   (so no worry about thi duplication)
-  ##############################################################################
-
+
   nullout <- list(theo = NULL, expt = NULL, ith = NULL, iex = NULL, m = NULL)
   len <- length(theos)
 
@@ -1267,7 +1265,7 @@ search_mgf <- function (expt_mass_ms1 = NULL, expt_moverz_ms2 = NULL,
   if (!by_modules)
     pep_mod_groups <- pep_mod_groups[oks2]
 
-  ans <- mapply(
+  mapply(
     function (x, y, g) {
       attr(x, "theo_ms1") <- y
       attr(x, "pep_mod_group") <- g
@@ -1320,8 +1318,6 @@ search_mgf <- function (expt_mass_ms1 = NULL, expt_moverz_ms2 = NULL,
   #   4  790.  790.
   #   5  868.  868.
   #   6 1297. 1297.
-
-  invisible(ans)
 }
 
 

diff --git a/R/msmsmatches.R b/R/msmsmatches.R
@@ -148,12 +148,8 @@
 #'   1} or \code{0}.
 #' @param maxn_vmods_sitescombi_per_pep A non-negative integer; the maximum
 #'   number of combinatorial variable modifications per peptide sequence (per
-#'   module). The combinations include the ways of neutral losses and
-#'   modification labels. For instance, at \code{maxn_vnl_per_seq = 2}, the ways
-#'   of allowed position permutation will be 32. Similarly, at
-#'   \code{maxn_vnl_per_seq = 2} and, for example, the permutation of labels
-#'   \code{Acetyl (K)} and \code{TMT (K)} (let's say three ways), the ways of
-#'   position permutation will be limited to \code{64/2/3}.
+#'   module). The ways include the permutations in neutral losses and
+#'   modifications (e.g., \code{Acetyl (K) and TMT (K)}).
 #' @param min_len A positive integer; the minimum length of peptide sequences
 #'   for considerations. Shorter peptides will be excluded. The default is 7.
 #' @param max_len A positive integer; the maximum length of peptide sequences
@@ -1534,7 +1530,7 @@ matchMS <- function (out_path = "~/mzion/outs",
   df <- dplyr::mutate(df, pep_expect = 10^((pep_score_co - pep_score)/10) * target_fdr)
   df[["pep_score_co"]] <- NULL
   df$pep_delta <- df$pep_exp_mr - df$pep_calc_mr
-  
+
   nms <- names(df)
   df  <- dplyr::bind_cols(
     df[grepl("^prot_", nms)],
@@ -1552,10 +1548,14 @@ matchMS <- function (out_path = "~/mzion/outs",
   rows_tmt <- grepl("TMT", df[["pep_fmod"]]) | grepl("TMT", df[["pep_vmod"]])
   df[!rows_tmt, cols_tmt] <- NA_real_
   rm(list = c("cols_tmt", "rows_tmt"))
-
-  readr::write_tsv(df, file.path(out_path, "psmC.txt"))
 
   local({
+    df$pep_exp_mr  <- round(df$pep_exp_mr, digits = 4L)
+    df$pep_calc_mr <- round(df$pep_calc_mr, digits = 4L)
+    df$pep_delta   <- round(df$pep_delta, digits = 4L)
+    df$pep_expect  <- format(df$pep_expect, digits = 3L)
+    readr::write_tsv(df, file.path(out_path, "psmC.txt"))
+
     session_info <- sessionInfo()
     save(session_info, file = file.path(out_path, "Calls", "mzion.rda"))
   })
@@ -1920,6 +1920,11 @@ post_psmC2Q <- function (df, dfC, tier = NULL)
     df[!grepl("^prot_|^pep_|^psm_", names(df))],
   )
 
+  df$pep_exp_mr  <- round(df$pep_exp_mr, digits = 4L)
+  df$pep_calc_mr <- round(df$pep_calc_mr, digits = 4L)
+  df$pep_delta   <- round(df$pep_delta, digits = 4L)
+  df$pep_expect  <- format(df$pep_expect, digits = 3L)
+
   df <- dplyr::select(df, -which(names(df) %in% c("prot_n_psm", "prot_n_pep")))
 }
 

diff --git a/R/scores.R b/R/scores.R
@@ -1135,9 +1135,10 @@ add_primatches <- function (file = NULL, tempdir = NULL, add_ms2theos = FALSE,
 {
   df <- qs::qread(file.path(tempdir, file))
 
+  if (!add_ms2moverzs) df[["pep_ms2_moverzs"]] <- NA_character_
+  if (!add_ms2ints)    df[["pep_ms2_ints"]] <- NA_character_
+
   df <- dplyr::mutate(df, 
-                      pep_ms2_moverzs = NA_character_, 
-                      pep_ms2_ints = NA_character_, 
                       pep_ms2_theos = NA_character_, 
                       pep_ms2_theos2 = NA_character_, 
 
@@ -1231,8 +1232,8 @@ add_primatches <- function (file = NULL, tempdir = NULL, add_ms2theos = FALSE,
 
   if (add_ms2theos) df$pep_ms2_theos <- collapse_vecs(lapply(pris, `[[`, "theo"))
   if (add_ms2theos2) df$pep_ms2_theos2 <- collapse_vecs(lapply(secs, `[[`, "theo"))
-  if (add_ms2moverzs) df$pep_ms2_moverzs <- collapse_vecs(df$ms2_moverz)
-  if (add_ms2ints) df$pep_ms2_ints <- collapse_vecs(df$ms2_int)
+  if (add_ms2moverzs) df$pep_ms2_moverzs <- collapse_vecs(df$pep_ms2_moverzs)
+  if (add_ms2ints) df$pep_ms2_ints <- collapse_vecs(df$pep_ms2_ints)
 
   qs::qsave(df, file.path(tempdir, gsub("^prescores", "ms2info", file)), 
             preset = "fast")
@@ -1248,6 +1249,8 @@ add_primatches <- function (file = NULL, tempdir = NULL, add_ms2theos = FALSE,
 #' @param sep A separator.
 collapse_vecs <- function (vecs, nm = "theo", sep = ";") 
 {
+  vecs <- lapply(vecs, round, digits = 4L)
+
   ans <- lapply(vecs, function (v) 
     .Internal(paste0(list(v), collapse = sep, recycle0 = FALSE))
   )

diff --git a/man/calc_pepscores.Rd b/man/calc_pepscores.Rd
diff --git a/man/calib_mgf.Rd b/man/calib_mgf.Rd
diff --git a/man/frames_adv.Rd b/man/frames_adv.Rd
diff --git a/man/gen_ms2ions_base.Rd b/man/gen_ms2ions_base.Rd
diff --git a/man/hms2match.Rd b/man/hms2match.Rd
diff --git a/man/matchMS.Rd b/man/matchMS.Rd
diff --git a/man/mframes_adv.Rd b/man/mframes_adv.Rd
diff --git a/man/ms2match.Rd b/man/ms2match.Rd
diff --git a/man/ms2match_all.Rd b/man/ms2match_all.Rd
diff --git a/man/ms2match_one.Rd b/man/ms2match_one.Rd