@@ -208,7 +208,10 @@ genotype <- function(filenames,

 	close(callSet)

 	return(callSet)

 }

-genotypeLD <- genotype2 <- genotype

+genotype2 <- function(){

+	.Defunct(msg="The genotype2 function has been deprecated. The function genotype should be used instead.  genotype will support large data using ff provided that the ff package is loaded.")

+}

+genotypeLD <- genotype2

 rowCovs <- function(x, y, ...){

 	notna <- !is.na(x)

@@ -321,34 +321,34 @@ setReplaceMethod("flags", signature=signature(object="CNSet", value="ff_matrix")

 ##	ellipse.CNSet(x, copynumber, batch, ...)

 ##})

-##ACN.X <- function(object, allele, i, j){

-##	acn <- list()

-##	batches <- unique(batch(object)[j])

-##	for(k in seq_along(batches)){

-##		l <- match(batches[k], bns)

-##		nuA <- nuA(object)[ii, l]

-##		nuB <- nuB(object)[ii, l]

-##		phiA <- phiA(object)[ii, l]

-##		phiB <- phiB(object)[ii, l]

-##		phiPrimeA <- phiPrimeA(object)[ii, l]

-##		phiPrimeB <- phiPrimeB(object)[ii, l]

-##		if(all(is.na(phiPrimeA))){

-##			I <- allele(object, allele)[ii, j]

-##			if(allele=="A") acn[[k]] <- 1/phiA*(I - nuA)

-##			if(allele=="B") acn[[k]] <- 1/phiB*(I - nuB)

-##		} else {

-##			A <- A(object)[ii, j]

-##			B <- B(object)[ii, j]

-##			phistar <- phiPrimeB/phiA

-##			tmp <- (B-nuB - phistar*A + phistar*nuA)/phiB

-##			cb <- tmp/(1-phistar*phiPrime/phiB)

-##			ca <- A-nuA-phiPrimeA*cb/phiA

-##			if(allele=="A") acn[[k]] <- ca

-##			if(allele=="B") acn[[k]] <- cb

-##		}

-##	}

-##	return(acn)

-##}

+ACN.X <- function(object, allele, i, j){

+	acn <- list()

+	batches <- unique(batch(object)[j])

+	for(k in seq_along(batches)){

+		l <- match(batches[k], bns)

+		nuA <- nuA(object)[ii, l]

+		nuB <- nuB(object)[ii, l]

+		phiA <- phiA(object)[ii, l]

+		phiB <- phiB(object)[ii, l]

+		phiPrimeA <- phiPrimeA(object)[ii, l]

+		phiPrimeB <- phiPrimeB(object)[ii, l]

+		if(all(is.na(phiPrimeA))){

+			I <- allele(object, allele)[ii, j]

+			if(allele=="A") acn[[k]] <- 1/phiA*(I - nuA)

+			if(allele=="B") acn[[k]] <- 1/phiB*(I - nuB)

+		} else {

+			A <- A(object)[ii, j]

+			B <- B(object)[ii, j]

+			phistar <- phiPrimeB/phiA

+			tmp <- (B-nuB - phistar*A + phistar*nuA)/phiB

+			cb <- tmp/(1-phistar*phiPrime/phiB)

+			ca <- A-nuA-phiPrimeA*cb/phiA

+			if(allele=="A") acn[[k]] <- ca

+			if(allele=="B") acn[[k]] <- cb

+		}

+	}

+	return(acn)

+}

 ACN <- function(object, allele, i , j){

@@ -139,13 +139,13 @@ to small to be used for copy number estimation). We wrap the function

 Sweaving after the initial batch job.

 <<genotype>>=

-gtSet <- checkExists("gtSet.assayData_matrix",

+gtSet.tmp <- checkExists("gtSet.assayData_matrix",

 		     .path=outdir,

 		     .FUN=genotype,

 		     filenames=celFiles[1:5],

 		     cdfName=cdfName,

 		     batch=batch[1:5])

-}

+rm(gtSet.tmp); gc()

 @

 A more memory efficient approach to preprocessing and genotyping is

@@ -177,9 +177,9 @@ if(!file.exists(file.path(outdir, "cnSet.rda"))){

 	gtSet <- checkExists("gtSet",

 			     .path=outdir,

 			     .FUN=genotypeLD,

-			     filenames=celFiles,

+			     filenames=celFiles[1:200],

 			     cdfName=cdfName,

-			     batch=batch)

+			     batch=batch[1:200])

 	class(calls(gtSet))

 }

 @

@@ -243,35 +243,46 @@ At nonpolymorphic loci, either the 'CA' or totalCopynumber method can

 be used to obtain total copy number.

 <<ca>>=

-cn.nonpolymorphic <- CA(obj, i=which(!isSnp(obj)))

+cn.nonpolymorphic <- CA(cnSet, i=which(!isSnp(obj)))

 @

 Total copy number at both polymorphic and nonpolymorphic loci:

 <<totalCopynumber>>=

 ##cn <- copyNumber(x)

-cn <- totalCopyNumber(x, sample(1:nrow(x), 1e4), 1:5)

+cn <- totalCopyNumber(cnSet, sample(1:nrow(cnSet), 1e4), 1:5)

 apply(cn, 2, median, na.rm=TRUE)

 @

-The above estimates can be plotted versus physical position.

+The above estimates can be plotted versus physical position. Accessors

+for physical position and chromosome are also provided. In the

+following codechunk we extract the position and chromosome for the

+first 10 markers in the \Robject{cnSet} object.

+

+<<fdAccessors>>=

+position(cnSet)[1:10]

+chromosome(cnSet)[1:10]

+@

 \section{The CNSet container}

 The objects returned by the \Rfunction{genotype} and

 \Rfunction{crlmmCopynumber} have assay data elements that are pointers

-to \Robject{ff} objects stored in the directory \Robject{outdir}.  The

+to \Robject{ff} objects stored in the directory \Robject{outdir}.  Had

+we not loaded the \Rpackage{ff} prior to running these functions, the

+\Robject{AssayData} elements would be ordinary matrices, though the

+RAM required for running the algorithm would be substantial. The

 functions \Rfunction{open} and \Rfunction{close} open and close the

 connections to the \Robject{assayData} elements. Subsetting an

-\Robject{ff} object pulls the data from disk into memory and should

-therefore be used with caution. In particular, subsetting the

-\Robject{gtSet} would subset each element in the \Robject{assayData}

-slot, returning an object of the same class but with

-\Robject{assayData} elements that are matrices.  Such an operation can

-be exceedingly slow when performed over a network and require

-subsantial RAM.  The preferred approach is to extract only the assay

-data element that is needed. In the example below, we extract the

-genotype calls for the the first 50 samples.

+\Robject{ff} object pulls the data from disk into memory and should be

+used with caution. In particular, subsetting the \Robject{gtSet} would

+subset each element in the \Robject{assayData} slot, returning an

+object of the same class but with \Robject{assayData} elements that

+are matrices.  Such an operation can be exceedingly slow when

+performed over a network and require subsantial RAM.  The preferred

+approach is to extract only the assay data element that is needed. In

+the example below, we extract the genotype calls for the the first 50

+samples.

 <<check>>=

 dims(cnSet)

@@ -295,8 +306,6 @@ slot can be listed as follows.

 assayDataElementNames(batchStatistics(cnSet))

 @

-

-

 Note that for the Affymetrix 6.0 platform the assay data elements each

 have a row dimension corresponding to the total number of polymorphic

 and nonpolymorphic markers interrogated by the Affymetrix 6.0

@@ -327,74 +336,43 @@ posterior.prob <- tryCatch(i2p(snpCallProbability(cnSet)),

 posterior.prob2 <- p2i(snpCallProbability(cnSet)[rows, cols])

 @

-Next, we obtain locus-level estimates of copy number.

-

-<<cn.ff, echo=FALSE, eval=FALSE>>=

-rm(cnSet); gc()

-ocProbesets(75e3)

-##ff objects

-system.time(cnSet2 <- crlmmCopynumber2(cnSet2))

-save(cnSet2, file=file.path(outdir, "cnSet2.rda"))

-@

-

-<<timings, eval=FALSE>>=

-tryCatch(print(paste("Time for matrix version:", time1)), error=function(e) print("timing for CN estimation not available"))

-tryCatch(print(paste("Time for ff version:", time2)), error=function(e) print("timing for CN estimation not available"))

-rm(cnSet2); gc()

-@

-

-

-\section{Accessors}

-

-For the remainder of this vignette, we illustrate accessors and

-visualizations using the sample object provided in the

-\Rpackage{crlmm} package.

-

-<<sampleset>>=

-data(sample.CNSetLM)

-x <- sample.CNSetLM

-@

-

-

-\subsection{Quantile-normalized intensities}

-

 Accessors for the quantile normalized intensities for the A allele at

 polymorphic loci:

 <<accessors>>=

-snp.index <- which(isSnp(x))

-np.index <- which(!isSnp(x))

-a <- (A(x))[snp.index, ]

+snp.index <- which(isSnp(cnSet))

+np.index <- which(!isSnp(cnSet))

+a <- (A(cnSet))[snp.index, ]

 dim(a)

 @

 The extra set of parentheses surrounding \Robject{A(cnSet2)} above is

 added to emphasize the appropriate order of operations. Subsetting the

-entire \Robject{x} object in the following code should be avoided for

-large datasets.

+entire \Robject{cnSet} object in the following, unevaluated codechunk

+should be avoided for large datasets.

 <<eval=FALSE>>=

-a <- A(x[snp.index, ])

+a <- A(cnSet[snp.index, ])

 @

-The quantile-normalized intensities for nonpolymorphic loci are obtained

-by:

+The quantile-normalized intensities for nonpolymorphic loci are

+obtained by:

 <<>>=

-npIntensities <- (A(x))[np.index, ]

+npIntensities <- (A(cnSet))[np.index, ]

 @

 Quantile normalized intensities for the B allele at polymorphic loci:

 <<B.polymorphic>>=

-b.snps <- (B(x))[snp.index, ]

+b.snps <- (B(cnSet))[snp.index, ]

 @

 Note that NAs are recorded in the 'B' assay data element for

 nonpolymorphic loci:

 <<B.NAs>>=

-all(is.na((B(x))[np.index, ]))

+all(is.na((B(cnSet))[np.index, ]))

 @

 <<clean, echo=FALSE>>=

@@ -405,11 +383,11 @@ rm(b.snps, a, npIntensities); gc()

 Genotype calls:

 <<genotypes>>=

-genotypes <- (snpCall(x))[snp.index, ]

+genotypes <- (snpCall(cnSet))[snp.index, ]

 @

 Confidence scores of the genotype calls:

 <<confidenceScores>>=

-genotypeConf <- integerScoreToProbability(snpCallProbability(x)[snp.index[1:10], ])

+genotypeConf <- integerScoreToProbability(snpCallProbability(cnSet)[snp.index[1:10], ])

 @

 \paragraph{\Robject{CopyNumberSet}: allele-specific copy number}

@@ -421,18 +399,18 @@ genotypeConf <- integerScoreToProbability(snpCallProbability(x)[snp.index[1:10],

 Information on physical position and chromosome can be accessed as follows:

 <<positionChromosome>>=

-xx <- position(x)

+xx <- position(cnSet)

 yy <- chromosome(x)

 @

 Parameters from the linear model used to estimate copy number are

-stored in the slot \Robject{lM}.

+stored in the slot \Robject{batchStatistics}.

 <<copynumberParameters>>=

-names(lM(x))

-dim(lM(x)[[1]])

+assayDataElementNames(batchStatistics(cnSet))

 @

+TODO: Describe accessors for batch-level summaries.

 \section{Suggested visualizations}

@@ -442,7 +420,7 @@ dim(lM(x)[[1]])

 A histogram of the signal to noise ratio for the HapMap samples:

 <<plotSnr, fig=TRUE, include=FALSE>>=

-hist(x$SNR, xlab="SNR", main="", breaks=25)

+hist(cnSet$SNR, xlab="SNR", main="", breaks=25)

 @

 \begin{figure}

@@ -463,13 +441,13 @@ area of research.

 <<oneSample, fig=TRUE, width=8, height=4, include=FALSE>>=

 par(las=1, mar=c(4, 5, 4, 2))

-plot(position(x), copyNumber(x)[, 1], pch=".",

+plot(position(cnSet), copyNumber(cnSet)[, 1], pch=".",

      cex=2, xaxt="n", col="grey20", ylim=c(0,4),

      ylab="copy number", xlab="physical position (Mb)",

-     main=paste(sampleNames(x)[1], ", CHR:", unique(chromosome(x))))

-points(position(x)[!isSnp(x)], copyNumber(x)[!isSnp(x), 1],

+     main=paste(sampleNames(cnSet)[1], ", CHR:", unique(chromosome(cnSet))))

+points(position(cnSet)[!isSnp(cnSet)], copyNumber(cnSet)[!isSnp(cnSet), 1],

        pch=".", cex=2, col="lightblue")

-axis(1, at=pretty(range(position(x))), labels=pretty(range(position(x)))/1e6)

+axis(1, at=pretty(range(position(cnSet))), labels=pretty(range(position(cnSet)))/1e6)

 @

 <<idiogram, eval=FALSE, echo=FALSE>>=

@@ -526,59 +504,6 @@ plotCytoband(22, new=FALSE, cytoband.ycoords=c(3.8, 4), label.cytoband=FALSE)

 Scatterplots of the A and B allele intensities (log-scale) can be

 useful for assessing the biallelic genotype calls.  This section of

 the vignette is currently under development.

-% The following code chunk is

-%displayed in Figure \ref{fig:prediction}.

-

-<<predictionRegions, fig=TRUE, width=8, height=8, include=FALSE, eval=FALSE>>=

-i <- snp.index[1]

-#plotCNSetLM=crlmm:::plotCNSetLM

-##trace(plotCNSetLM, browser)

-plot(i, x, copynumber=2)

-##myScatter <- function(object, add=FALSE, ...){

-##	A <- log2(A(object))

-##	B <- log2(B(object))

-##	if(!add){

-##		plot(A, B, ...)

-##	} else{

-##		points(A, B, ...)

-##	}

-##}

-##index <- which(isSnp(cnSet))[1:9]

-##xlim <- ylim <- c(6.5,13)

-##par(mfrow=c(3,3), las=1, pty="s", ask=FALSE, mar=c(2, 2, 2, 2), oma=c(2, 2, 1, 1))

-##for(i in index){

-##	gt <- calls(cnSet)[i, ]

-##	if(i != 89){

-##		myScatter(cnSet[i, ],

-##			  pch=pch,

-##			  col=colors[snpCall(cnSet)[i, ]],

-##			  bg=colors[snpCall(cnSet)[i, ]], cex=cex,

-##			  xlim=xlim, ylim=ylim)

-##		mtext("A", 1, outer=TRUE, line=1)

-##		mtext("B", 2, outer=TRUE, line=1)

-##		crlmm:::ellipse.CNSet(cnSet[i, ], copynumber=2, batch="C", lwd=2, col="black")

-##		crlmm:::ellipse.CNSet(cnSet[i, ], copynumber=2, batch="Y", lwd=2, col="grey50")

-##	} else {

-##		plot(0:1, xlim=c(0,1), ylim=c(0,1), type="n", xaxt="n", yaxt="n")

-##		legend("center",

-##		       legend=c("CN = 2, CEPH", "CN = 2, Yoruban"),

-##		       col=c("black", "grey50"), lwd=2, bty="n")

-##	}

-##}

-@

-

-%\begin{figure}

-%  \centering

-%  \includegraphics[width=0.8\textwidth]{copynumber-predictionRegions}

-%  \caption{\label{fig:prediction} Scatterplots of A versus B

-%    intensities.  Each panel displays a single SNP. The ellipses

-%    indicate the 95\% probability region for copy number 2 for the CEPH

-%    (black) and Yoruban subjects (grey).}

-%\end{figure}

-

-%\section{Details for the copy number estimation procedure}

-%

-%See the technical report \citep{Scharpf2009}.

 \section{Session information}

 <<sessionInfo, results=tex>>=

@@ -98,6 +98,11 @@ crlmmCopynumber(object, MIN.SAMPLES=10, SNRMin = 5, MIN.OBS = 1,

     \code{MIN.NU} and \code{MIN.PHI} are ignored. When TRUE, background

     (nu) and slope (phi) coefficients below MIN.NU and MIN.PHI are set

     to MIN.NU and MIN.PHI, respectively.}

+

+  \item{type}{ Character string vector that must be one or more of

+    "SNP", "NP", "X.SNP", or "X.NP". Type refers to a set of

+    markers. See details below}

+    

  }

  \references{

@@ -133,6 +138,19 @@ crlmmCopynumber(object, MIN.SAMPLES=10, SNRMin = 5, MIN.OBS = 1,

 	The functions \code{crlmmCopynumberLD} and

 	\code{crlmmCopynumber2} have been deprecated.

+	The argument \code{type} can be used to specify a subset of

+	markers for which the copy number estimation algorithm is run.

+	One or more of the following possible entries are valid: 'SNP',

+	'NP', 'X.SNP', and 'X.NP'.

+

+	'SNP' referers to autosomal SNPs.

+

+	'NP' refers to autosomal nonpolymorphic markers.

+

+	'X.SNP' refers to SNPs on chromosome X.

+

+	'X.NP' refers to autosomes on chromosome X.

+

 }

 \author{R. Scharpf}

 \keyword{manip}

@@ -31,28 +31,23 @@ data(sample.CNSetLM)

 cnSet <- as(sample.CNSetLM, "CNSet")

 all(isCurrent(cnSet)) ## is the cnSet object current?

-## updating class CNSetLM using ff objects

-library(ff)

-path <- system.file("extdata", package="crlmm")

-ldPath(path)

-data(sample.CNSetLMff)

-cnSetff <- as(sample.CNSetLMff, "CNSet")

-all(isCurrent(cnSet))

-

-## a bigger object with multiple batches

 \dontrun{

-	outdir <- "/amber1/scratch/rscharpf/jss/hapmap2"

-	load(file.path(outdir, "container.rda"))

-	container <- object; rm(object); gc()

-	container2 <- as(container, "CNSet")

-	all(isCurrent(container2))

-	## test replacement methods, subset methods

-	table(batch(container2))

-	##generates warning ... would need open, close in the '[' method

-	invisible(open(nuA(container2)))

-	xx <- nu(container2, "A")[1:5, ]

-	nuA(container2)[1:5, ] <- xx

-	invisible(close(nuA(container2)))

+	## updating class CNSetLM using ff objects

+	## a bigger object with multiple batches

+	if(require(ff)){

+		outdir <- "/amber1/scratch/rscharpf/jss/hapmap2"

+		load(file.path(outdir, "container.rda"))

+		container <- object; rm(object); gc()

+		container2 <- as(container, "CNSet")

+		all(isCurrent(container2))

+		## test replacement methods, subset methods

+		table(batch(container2))

+		##generates warning ... would need open, close in the '[' method

+		invisible(open(nuA(container2)))

+		xx <- nu(container2, "A")[1:5, ]

+		nuA(container2)[1:5, ] <- xx

+		invisible(close(nuA(container2)))

+	}

 }

 ## --------------------------------------------------

 ## accessors for the feature-level info