add utils functions and new as(mzRident, "data.frame")

Author: LaurentGatto

Diff for: DESCRIPTION

@@ -48,6 +48,7 @@ Depends:
     ProtGenerics (>= 1.5.1)
 Imports:
     plyr,
+    dplyr,
     IRanges,
     preprocessCore,
     vsn,
@@ -57,7 +58,7 @@ Imports:
     impute,
     pcaMethods,
     MALDIquant (>= 1.16),
-    mzID (>= 1.5.2),    
+    mzID (>= 1.5.2),
     digest,
     lattice,
     ggplot2,

Diff for: NAMESPACE

@@ -38,11 +38,12 @@ importFrom(S4Vectors, isEmpty, metadata)
 ## importFrom(rols, olsQuery, term)
 ## importClassesFrom(rols, CVParam)
 importFrom(pcaMethods, pca, completeObs)
-importFrom(plyr, ddply, llply, adply, summarise, summarize)
+importFrom(plyr, ddply, llply)
+importFrom(dplyr, left_join)
 importFrom(stats4, plot)
 importFrom(preprocessCore, normalize.quantiles)
 importFrom(vsn, vsn2, meanSdPlot)
-importFrom(IRanges, IRanges)
+importFrom(IRanges, IRanges, reduce)
 importClassesFrom(IRanges, IRanges)
 importFrom(affy, MAplot, ma.plot, mva.pairs)
 importFrom(mzID, mzID, flatten)
@@ -67,9 +68,9 @@ export(MSnSet,
        ## extractSpectra, defunct in v 1.5.3
        extractPrecSpectra,
        as.matrix.FoICollection,
+       as.data.frame.mzRident,
        as.data.frame.Spectrum,
        as.data.frame.MSnSet, ms2df,
-       as.data.frame.mzRident,       
        as.ExpressionSet.MSnSet,
        as.MSnSet.ExpressionSet,
        as.MIAME.MIAPE,
@@ -130,7 +131,8 @@ export(MSnSet,
        productMz,
        aggregationFun,
        Chromatogram,
-       Chromatograms
+       Chromatograms,
+       factorsAsStrings
        )
 
 exportClasses(pSet,
@@ -277,7 +279,8 @@ exportMethods(updateObject,
               pData, "pData<-",
               "$",
               "$<-",
-              chromatogram)
+              chromatogram,
+              reduce)
 
 ## methods NOT exported
 ## curveStats

Diff for: NEWS.md

@@ -1,7 +1,10 @@
 # MSnbase 2.3
 
 ## Changes in version 2.3.9
-- Nothing yet
+- Using new mzR::openIdfile backend to add identifcation data to raw
+  and quantitative data (see issue #232) <2017-07-28 Fri>
+- New utils functions: factorsAsStrings, makeCamelCase and
+  reduce,data.frame <2017-07-29 Sat>
 
 ## Changes in version 2.3.8
 - New infrastructure for chromatogram data <2017-06-24 Sat>

Diff for: R/functions-mzR.R

@@ -111,33 +111,48 @@ plotMzDelta_list <- function(object,            ## peakLists
     invisible(p)
 }
 
+##' A function to convert the identification data contained in an
+##' \code{mzRident} object to a \code{data.frame}. Each row represents
+##' a scan, which can however be repeated several times if the PSM
+##' matches multiple proteins and/or contains two or more
+##' modifications. To reduce the \code{data.frame} so that rows/scans
+##' are unique, use \link{\code{reduce}}.
+##' 
+##' @title Coerce identification data to a \code{data.frame}
+##' @param from An object of class \code{mzRident} defined in the
+##'     \code{mzR} package.
+##' @return A \code{data.frame}
+##' @author Laurent Gatto
+##' @name as
+##' @rdname mzRident2dfr
+##' @examples
+##' ## find path to a mzIdentML file
+##' identFile <- dir(system.file(package = "MSnbase", dir = "extdata"),
+##'                  full.name = TRUE, pattern = "dummyiTRAQ.mzid")
+##' library("mzR")
+##' x <- openIDfile(identFile)
+##' x
+##' as(x, "data.frame")
 setAs("mzRident", "data.frame",
       function(from) {
-          iddf <- psms(from)
+          ## peptide spectrum matching
+          iddf <- factorsAsStrings(psms(from))
+          ## add file raw and mzid provenances      
           iddf$spectrumFile <- basename(sourceInfo(from))
           iddf$idFile <- basename(fileName(from))
-          iddf <- utils.leftJoin(iddf, score(from),
-                                 by.x = "spectrumID",
-                                 by.y = "spectrumID")
-          mods <- modifications(from)
-          names(mods)[-1] <- paste0("mod.", names(mods)[-1])
-          names(mods)[-1] <- gsub('\\.(\\w?)', '\\U\\1', names(mods)[-1], perl = TRUE)
-          iddf <- utils.leftJoin(iddf, mods,
-                                 by.x = "spectrumID",
-                                 by.y = "spectrumID")
-          subs <- substitutions(from)
-          names(subs)[-1] <- paste0("sub.", names(subs)[-1])
-          names(subs)[-1] <- gsub('\\.(\\w?)', '\\U\\1', names(subs)[-1], perl = TRUE)
-          iddf <- utils.leftJoin(iddf, subs,
-                                 by.x = "spectrumID",
-                                 by.y = "spectrumID")
-          iddf <- lapply(iddf,
-                         function(x) {
-                             if (is.factor(x)) as.character(x)
-                             else x
-                         })
-          data.frame(iddf, stringsAsFactors = FALSE)
+          ## add scores
+          scores <- factorsAsStrings(score(from))
+          iddf <- suppressMessages(left_join(iddf, scores))
+          ## add modification
+          mods <- factorsAsStrings(modifications(from))
+          names(mods)[-1] <- makeCamelCase(names(mods), prefix = "mod")[-1]
+          iddf <- suppressMessages(left_join(iddf, mods))
+          ## add substitutions 
+          subs <- factorsAsStrings(substitutions(from))
+          names(subs)[-1] <- makeCamelCase(names(subs), prefix = "sub")[-1]
+          suppressMessages(left_join(iddf, subs))
       })
 
 as.data.frame.mzRident <-
     function(x, row.names = NULL, optional = FALSE, ...) as(x, "data.frame")
+

Diff for: R/methods-MSnSet.R

@@ -656,10 +656,12 @@ setMethod("addIdentificationData", c("MSnSet", "character"),
                    verbose = isMSnbaseVerbose()) {
               if (length(id) == 1 && file.exists(id)) {
                   id <- mzR::openIDfile(id)
+                  id <- reduce(id, key = "spectrumID")
               } else {
                   if (!all(flex <- file.exists(id)))
                       stop(paste(id[!flex], collapse = ", "), " not found.")
                   id <- lapply(id, function(x) as(openIDfile(x), "data.frame"))
+                  id <- lapply(id, reduce, key = "spectrumID")
                   id <- do.call(rbind, id)
               }
               addIdentificationData(object, id = id,
@@ -676,6 +678,7 @@ setMethod("addIdentificationData", c("MSnSet", "mzRident"),
                    pepseq = "sequence",
                    ...) {
               iddf <- as(id, "data.frame")
+              iddf <- reduce(iddf)
               addIdentificationData(object, iddf,
                                     fcol = fcol, icol = icol,
                                     acc, desc, pepseq)