additionally store mslists without formula annotation cleanup and background MSMS peaks

Author: rickhelmus

workflow.R

@@ -11,11 +11,11 @@ for (f in list.files(c("workflow", "utils"), pattern = "\\.R$", full.names = TRU
 wfData <- list()
 wfData$featData <- getFeatures()
 
-bgMSMS <- getBGMSMSPeaks(analysisInfo(wfData$featData$fGroupsRaw)[analysisInfo(wfData$featData$fGroupsRaw)$group == "MQ", ])
+wfData$bgMSMS <- getBGMSMSPeaks(analysisInfo(wfData$featData$fGroupsRaw)[analysisInfo(wfData$featData$fGroupsRaw)$group == "MQ", ])
 
-wfData$SuSData <- getSuspectsFromStructures(wfData$featData, bgMSMS)
-wfData$SuFData <- getSuspectsFromFormulas(wfData$featData, wfData$SuSData, bgMSMS)
-wfData$unkData <- getUnknownsGeneral(wfData$featData, wfData$SuSData, wfData$SuFData, bgMSMS)
+wfData$SuSData <- getSuspectsFromStructures(wfData$featData, wfData$bgMSMS)
+wfData$SuFData <- getSuspectsFromFormulas(wfData$featData, wfData$SuSData, wfData$bgMSMS)
+wfData$unkData <- getUnknownsGeneral(wfData$featData, wfData$SuSData, wfData$SuFData, wfData$bgMSMS)
 wfData$unkFormData <- getUnknownsFormulas(wfData$unkData, wfData$SuSData)
 wfData$unkCompData <- getUnknownsCompounds(wfData$unkData, wfData$SuSData)
 wfData$confData <- getIDLConfirmations()

workflow/suspects-formulas.R

@@ -29,7 +29,7 @@ getSuspectsFromFormulas <- function(featData, SuSData, bgMSMS)
     fGroups <- doScreenSuspects(featData$fGroupsSel, suspListTPs, onlyHits = TRUE)
 
     # the next steps largely follow that of the structure suspects annotation
-    mslists <- getMSPL(fGroups, bgMSMS)
+    mslists <- mslistsNoAnnF <- getMSPL(fGroups, bgMSMS)
     formulas <- getFormulas(fGroups, mslists)
     formulas <- filterFormulasForSusps(fGroups, formulas)
     mslists <- filterMSPLForAnn(mslists, fGroups)
@@ -45,6 +45,7 @@ getSuspectsFromFormulas <- function(featData, SuSData, bgMSMS)
         TPsFormF = TPsFormF,
         fGroups = fGroups,
         mslists = mslists,
+        mslistsNoAnnF = mslistsNoAnnF,
         formulas = formulas,
         componentsTPs = componentsTPs
     ))

workflow/suspects-structures.R

@@ -20,7 +20,7 @@ getSuspectsFromStructures <- function(featData, bgMSMS)
     fGroups <- doScreenSuspects(featData$fGroupsSel, suspListTPs, onlyHits = TRUE)
 
     # formula/compound annotation for structure suspects
-    mslists <- getMSPL(fGroups, bgMSMS)
+    mslists <- mslistsNoAnnF <- getMSPL(fGroups, bgMSMS)
     formulas <- getFormulas(fGroups, mslists)
     formulas <- filterFormulasForSusps(fGroups, formulas)
     # make customized compound database for MetFrag, so we don't rely on the presence of the TP suspects in public databases
@@ -94,6 +94,7 @@ getSuspectsFromStructures <- function(featData, bgMSMS)
         TPsCTS = TPsCTS, TPsBTE = TPsBTE, TPsBTH = TPsBTH, TPsPC = TPsPC, TPsLIT = TPsLIT, TPsCons = TPsCons,
         fGroups = fGroups,
         mslists = mslists,
+        mslistsNoAnnF = mslistsNoAnnF,
         formulas = formulas,
         compounds = compounds,
         componentsTPs = componentsTPsF

workflow/unknowns-general.R

@@ -7,7 +7,7 @@ getUnknownsGeneral <- function(featData, SuSData, SuFData, bgMSMS)
                                setdiff(names(fGroups), union(groupNames(SuSData$componentsTPs),
                                                              groupNames(SuFData$componentsTPs))))]
 
-    mslists <- getMSPL(fGroups, bgMSMS)
+    mslists <- mslistsNoAnnF <- getMSPL(fGroups, bgMSMS)
 
     componentsTPs <- generateComponents(fGroups[, suspects = getParentSuspList()$name], "tp", fGroupsTPs = fGroups,
                                         MSPeakLists = mslists, ignoreParents = TRUE)
@@ -37,6 +37,7 @@ getUnknownsGeneral <- function(featData, SuSData, SuFData, bgMSMS)
         fGroups = fGroups,
         componentsTPs = componentsTPs,
         mslists = mslists,
+        mslistsNoAnnF = mslistsNoAnnF,
         formulas = formulas,
         compoundsUF = compoundsUF, compounds = compounds
     ))