From 351f363283586bf7df2491c22e4f730aa16df7e7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Todor=20Kondi=C4=87?= <kontrapunkt@uclmail.net>
Date: Mon, 16 Jan 2023 14:51:59 +0100
Subject: [PATCH] metfrag: summarise_metfrag_results: Update. errors:
check_key_absent: New error function.
---
DESCRIPTION | 2 +-
R/errors.R | 7 ++++
R/metfrag.R | 61 +++++++++++++++++++++++++++++++----
R/resources.R | 19 +++++++----
tests/testthat/test-metfrag.R | 53 +++---------------------------
5 files changed, 79 insertions(+), 63 deletions(-)
diff --git a/DESCRIPTION b/DESCRIPTION
index faf7087..2059e21 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,6 +1,6 @@
Package: shinyscreen
Title: Pre-screening of Mass Spectrometry Data
-Version: 1.2.9004
+Version: 1.2.9005
Author: Todor Kondić
Maintainer: Todor Kondić <todor.kondic@uni.lu>
Authors@R:
diff --git a/R/errors.R b/R/errors.R
index 53f048e..2859b70 100644
--- a/R/errors.R
+++ b/R/errors.R
@@ -37,3 +37,10 @@ check_extension <- function(extfileval,what) {
if (extfileval[[1]]==extfileval[[2]]) stop(errorCondition(paste0("We could not find the extension for ",what, ". The returned value was: ", extfileval[[2]]),
class = paste0(what,'-no-ext-found')))
}
+
+
+check_key_absent <- function(key,l,what) {
+ nms = names(l)
+ if (! (key %in% nms)) stop(errorCondition(paste0("Key",key, " has not been found for ", what),
+ class = paste0(what,'-absent')))
+}
diff --git a/R/metfrag.R b/R/metfrag.R
index 8d8ae9b..747ae2c 100644
--- a/R/metfrag.R
+++ b/R/metfrag.R
@@ -41,6 +41,12 @@ metfrag_get_ms2_spec <- function(ms2,stag_entry) {
x
}
+get_mf_res_ext <- function(fn) {
+ ext = sub(pattern = r"(^.*\.([[:alnum:]]+)$)",r"(\1)", fn)
+ check_extension(c(ext=ext,file=fn),what="mf-res-file")
+ ext
+}
+
metfrag_run <- function(param,path,subpaths,db_path,stag_tab,ms2,runtime,java_bin,nproc = 1L) {
keys = intersect(colnames(stag_tab),colnames(ms2))
message("Generating MetFrag configs.")
@@ -69,8 +75,7 @@ metfrag_run <- function(param,path,subpaths,db_path,stag_tab,ms2,runtime,java_bi
pth = file.path(path,subpaths[["results"]])
a_res_f = list.files(path = pth,
pattern = param$SampleName)[[1]]
- ext = sub(pattern = r"(^.*\.([[:alnum:]]+)$)",r"(\1)", a_res_f)
- check_extension(c(ext=ext,file=a_res_f),what="mf-res-file")
+ ext = get_mf_res_ext(a_res_f)
file_tab[,f_res:=paste0(param$SampleName,"_",stag,".",(ext))]
@@ -93,10 +98,6 @@ get_metfrag_targets <- function(stag_tab,ms2) {
}
-## get_metfrag_run_data(stag_tab,ms2,n=1) {
-## }
-
-
metfrag_on_state <- function(mconf,mrun,summ) {
@@ -212,5 +213,51 @@ metfrag_run_many <- function(fn_jar,file_tab, mem = NA_character_, java_bin = "j
}
-summarise_metfrag_results <- function(m) {
+summarise_metfrag_results <- function(param,path,subpaths,file_tab) {
+
+ ## which(max(as.numeric(mf_res$Score))==as.numeric(mf_res$Score))
+ index_maxScore = 1L
+
+ ## First detect which reader we need.
+ ext = get_mf_res_ext(file_tab[1,f_res])
+ readf = METFRAG_RESULT_READF[[ext]]
+ keyz = as.character(union(key(file_tab),"stag"))
+ .read_results <- function() {
+ file_tab[,{
+ fn = file.path(..path,subpaths$results,f_res)
+ dt = data.table::rbindlist(lapply(fn,function (ff) as.data.table(readf(ff))))
+ dt
+ },
+ by = keyz]
+ }
+
+ .adapt_col_types <- function(x) {
+ x[,c("Score",
+ "NoExplPeaks",
+ "NumberPeaksUsed",
+ "FragmenterScore",
+ "OfflineIndividualMoNAScore"):=lapply(.SD, as.numeric),.SDcol=c("Score",
+ "NoExplPeaks",
+ "NumberPeaksUsed",
+ "FragmenterScore",
+ "OfflineIndividualMoNAScore")]
+ }
+
+ .calc_basic_scores <- function(x) {
+ x[,.(num_poss_IDs=length(Score),
+ max_Score=max(Score),
+ n_Score_GE4=length(which(Score>=4)),
+ n_Score_GE3=length(which(Score>=3)),
+ n_Score_GE2=length(which(Score>=2))),
+ keyby="stag"]
+
+ }
+
+ thetab = .read_results()
+ thetab = .adapt_col_types(thetab)
+ monatab = thetab[,.(max_MoNAIndiv=max(OfflineIndividualMoNAScore))]
+ basictab = .calc_basic_scores(thetab)
+
+
}
+
diff --git a/R/resources.R b/R/resources.R
index 290ca67..b00f15b 100644
--- a/R/resources.R
+++ b/R/resources.R
@@ -343,16 +343,21 @@ METFRAG_PREPFLT_CHOICES = c("UnconnectedCompoundFilter","IsotopeFilter")
METFRAG_PREPFLT_DEFAULT = c("UnconnectedCompoundFilter","IsotopeFilter")
METFRAG_POSTPFLT_CHOICES = c("InChIKeyFilter")
METFRAG_POSTPFLT_DEFAULT = c("InChIKeyFilter")
-METFRAG_DEFAULT_SCORES = "FragmenterScore,OfflineIndividualMoNAScore" # AutomatedPeakFingerprintAnnotationScore,
- # AutomatedLossFingerprintAnnotationScore
- # supposedly
- # do
- # not
- # work
- # well.
+METFRAG_DEFAULT_SCORES = c(FragmenterScore="FragmenterScore",
+ OfflineIndividualMoNAScore="OfflineIndividualMoNAScore")
+ # AutomatedPeakFingerprintAnnotationScore,
+ # AutomatedLossFingerprintAnnotationScore
+ # supposedly
+ # do
+ # not
+ # work
+ # well.
METFRAG_DEFAULT_WEIGHTS = "1,1"
METFRAG_DEFAULT_MAX_TREE_DEPTH = 2
METFRAG_SAMPLE_NAME = "sample"
METFRAG_DEFAULT_ABSMASSDEV = 1.E-3
METFRAG_DEFAULT_RELMASSDEV = 5
METFRAG_DB_SEARCH_RELDEV = 5
+
+METFRAG_RESULT_READF = list(csv = function(file,...) data.table::fread(file=file,...),
+ xml = function(file,...) readxl::read_excel(path=file,...))
diff --git a/tests/testthat/test-metfrag.R b/tests/testthat/test-metfrag.R
index d05ef02..a2baace 100644
--- a/tests/testthat/test-metfrag.R
+++ b/tests/testthat/test-metfrag.R
@@ -71,54 +71,11 @@ ok_return_val("metfrag_run",{
f)))
}
+ x = summarise_metfrag_results(param = m$conf$metfrag$param,
+ path = m$run$metfrag$path,
+ subpaths = m$run$metfrag$subpaths,
+ file_tab = ftab)
+
})
})
-
-## ok_return_val("metfrag_on_state",{
-## ## skip_if_not(file.exists(Sys.getenv("METFRAG_JAR")),"Environment variable METFRAG_JAR does not contain a path to MetFrag jar package.")
-## ## m = make_dummy_mf_project()
-## ## res = metfrag_on_state(m)
-## ## withr::with_dir(m$run$metfrag$path,{
-## ## fc_i = readChar(res$targets[1,files],nchars=file.size(res$targets[1,files]))
-## ## fc_i = gsub(paste0("LocalDatabasePath = ",m$run$metfrag$db_path),"",fc_i)
-## ## expect_snapshot(fc_i)
-
-## ## fs_i = readChar(res$targets[3,files],nchars=file.size(res$targets[3,files]))
-## ## fs_i = gsub(paste0("LocalDatabasePath = ",m$run$metfrag$db_path),"",fs_i)
-## ## expect_snapshot(fs_i)
-
-## ## fc_l = readChar(res$targets[.N-2,files],nchars=file.size(res$targets[.N-2,files]))
-## ## fc_l = gsub(paste0("LocalDatabasePath = ",m$run$metfrag$db_path),"",fc_l)
-## ## expect_snapshot(fc_l)
-
-## ## fs_l = readChar(res$targets[.N,files],nchars=file.size(res$targets[.N,files]))
-## ## fs_l = gsub(paste0("LocalDatabasePath = ",m$run$metfrag$db_path),"",fs_l)
-## ## expect_snapshot(fs_l)
-## ## })
-## })
-
-
-
-
-## test_that("MetFrag example works.",{
-## skip_if_not(file.exists(Sys.getenv("METFRAG_JAR")),"Environment variable METFRAG_JAR does not contain a path to MetFrag jar package.")
-## skip_if_offline()
-## withr::with_tempdir({
-## runtime = path.expand(Sys.getenv("METFRAG_JAR"))
-## fn_conf = system.file("testdata/example_parameter_file.txt",package = "shinyscreen")
-## fn_peaks = system.file("testdata/example_data.txt",package = "shinyscreen")
-## fn_log = "metfrag.log"
-## file.copy(fn_conf,basename(fn_conf))
-## file.copy(fn_peaks,basename(fn_peaks))
-
-## metfrag_run(fn_jar = runtime,
-## fn_conf = basename(fn_conf),
-## fn_log = fn_log)
-
-## content = readChar(fn_log,nchars=file.size(fn_log))
-## expect_true(grepl(r"(0 candidate\(s\) discarded during processing due to errors)",content))
-## })
-## })
-
-
--
GitLab