From 1257b85a7efd3329dd2b4810b1dbe75bef70a6c9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Todor=20Kondi=C4=87?= <kontrapunkt@uclmail.net>
Date: Fri, 3 Mar 2023 08:01:58 +0100
Subject: [PATCH] Remove main bits of setid table processing.
---
R/api.R | 10 +++-------
R/mix.R | 17 -----------------
2 files changed, 3 insertions(+), 24 deletions(-)
diff --git a/R/api.R b/R/api.R
index 5e90ae1..8b37188 100644
--- a/R/api.R
+++ b/R/api.R
@@ -151,8 +151,6 @@ load_compound_input <- function(m) {
cmpds[,("known"):=.(the_ifelse(!is.na(SMILES),"structure",the_ifelse(!is.na(Formula),"formula","mz")))]
m$input$tab$cmpds <- cmpds
fn_setid <- file.path(m$run$paths$project,m$conf$compounds$sets)
- m$input$tab$setid <- read_setid(fn_setid,
- m$input$tab$cmpds)
m
}
@@ -190,18 +188,16 @@ load_inputs <- function(m) {
mk_comp_tab <- function(m) {
message("Stage: comptab")
- setid <- m$input$tab$setid
- setkey(setid,set)
mzml<- m$input$tab$mzml
setkey(mzml,set)
cmpds<-m$input$tab$cmpds
- setkey(cmpds,ID)
+ setkey(cmpds,set,ID)
assert(nrow(cmpds)>0,msg="No compound lists have been provided.")
- assert(all(mzml[,unique(set)] %in% setid[,unique(set)]),msg="Not all set names in the `datatab' data file table match those in the provided set list.")
+ assert(all(mzml[,unique(set)] %in% cmpds[,unique(set)]),msg="Not all set names in the `datatab' data file table match those in the provided set list.")
assert(all(mzml[,!is.na(unique(adduct))]),msg="Some data file entries do not have selected adducts.")
message("Begin generation of the comprehensive table.")
- comp <- cmpds[setid,on="ID"][mzml,.(tag,adduct,ID,RT,set,Name,file,SMILES,Formula,mz,known),on="set",allow.cartesian=T]
+ comp <- cmpds[mzml,.(tag,adduct,ID,RT,set,Name,file,SMILES,Formula,mz,known),on="set",allow.cartesian=T]
tab2file(tab=comp,file=paste0("setidmerge",".csv"))
setkey(comp,known,set,ID)
message("Merged all sets.")
diff --git a/R/mix.R b/R/mix.R
index aa88970..5adef45 100644
--- a/R/mix.R
+++ b/R/mix.R
@@ -489,23 +489,6 @@ vald_comp_tab<-function(df,ndf,checkSMILES=F,checkMz=F,checkNames=F) {
df
}
-read_setid <- function(fn,cmpds) {
- assert(file.exists(fn),msg=paste("Please provide valid compounds set table:", fn))
- assert(nrow(cmpds) > 0,msg="Please provide at least one compounds list.")
- setid <- file2tab(fn,colClasses=c(ID="character"))
- x<-cmpds[setid,on='ID'][,.SD,.SDcols=c(colnames(setid),'known')]
-
- sids <- unique(setid$ID)
- cids <- unique(cmpds$ID)
- diff <- setdiff(sids,cids)
- assert(length(diff)==0,msg=paste("The following IDs from set table have not been found in the compound table:","------",print_table(dtable(diff)),"------",sep = "\n"))
- x
-}
-
-
-
-
-
verify_cmpd_l <- function(dt,fn) {
fields <- colnames(EMPTY_CMPD_LIST)
dtflds <- colnames(dt)
--
GitLab