@@ -1,5 +1,5 @@

 Package: preprocessCore

-Version: 0.99.11

+Version: 0.99.12

 Title: A collection of pre-processing functions

 Author: Benjamin Milo Bolstad <bmb@bmbolstad.com>

 Maintainer: Benjamin Milo Bolstad <bmb@bmbolstad.com>

@@ -44,7 +44,8 @@ normalize.quantiles <- function(x,copy=TRUE){

   #matrix(.C("qnorm_c", as.double(as.vector(x)), as.integer(rows), as.integer(cols))[[1]], rows, cols)

-  .Call("R_qnorm_c",x,copy, PACKAGE="preprocessCore");

+##  .Call("R_qnorm_c",x,copy, PACKAGE="preprocessCore");

+  .Call("R_qnorm_c_handleNA",x,copy, PACKAGE="preprocessCore");

 }

@@ -12,14 +12,19 @@

   \item{x}{A matrix of intensities where each column corresponds to a

     chip and each row is a probe.}

   \item{copy}{Make a copy of matrix before normalizing. Usually safer to

-    work with a copy}

+    work with a copy, but in certain situations not making a copy of the

+    matrix, but instead normalizing it in place will be more memory friendly.}

 }

 \details{This method is based upon the concept of a quantile-quantile

   plot extended to n dimensions. No special allowances are made for

   outliers. If you make use of quantile normalization 

   please cite Bolstad et al, Bioinformatics (2003).

-  Note that this function does not handle missing (ie NA) values.

+  This functions will handle missing data (ie NA values), based on the

+  assumption that the data is missing at random.

+

+  Note that the current implementation optimizes for better memory usage

+  at the cost of some additional run-time.

 }

@@ -28,6 +28,10 @@

   \code{\link[affy]{rma}} or \code{\link[affy]{expresso}}

   please cite Bolstad et al, Bioinformatics (2003).

+

+  These functions will handle missing data (ie NA values), based on the

+  assumption that the data is missing at random.

+  

 }

 \value{

@@ -54,10 +54,22 @@

  ** Nov 13, 2006 - remove median code

  ** May 20, 2007 - move to preprocessCore. clean up code.

  ** May 26, 2007 - fix memory leak in qnorm_c_determine_target

- ** Jul 12, 2007 - improved ties handling (fixes off by "half" error which affects odd numbers of ties)

+ ** Jul 12, 2007 - improved ties handling (fixes off by "half" error which affects even numbers of ties)

+ ** Jul 14, 2007 - add NA handling to qnorm_c_using_target and qnorm_c_determine_target

  **

  ***********************************************************/

+/*****************************************************************************************************

+ *****************************************************************************************************

+ **

+ ** GENERAL NOTE: Many of the functions take pointers for arguements that are essentially just

+ **               int's. This is mostly legacy for when the functions were called via .C() in R rather

+ **               than via the .Call() interface.

+ **

+ *****************************************************************************************************

+ *****************************************************************************************************/

+

+

 #include <stdio.h>

 #include <stdlib.h>

 #include <math.h>

@@ -1179,7 +1191,6 @@ SEXP R_qnorm_robust_weights(SEXP X, SEXP remove_extreme, SEXP n_remove){

  **                  normalized to)

  ** int *targetrows - length of target distribution vector

  **

- ** Note that it is assumed that there is no NA or Inf type values in the vectors. (ie no missing data)

  **

  ** if targetrows == rows then the standard methodology is used.

  ** 

@@ -1591,6 +1602,77 @@ SEXP R_qnorm_determine_target(SEXP X, SEXP targetlength){

+

+/*********************************************************

+ **

+ ** void qnorm_c_handleNA(double *data, int *rows, int *cols)

+ **

+ **  this is the function that actually implements the

+ ** quantile normalization algorithm. It is called from R.

+ **

+ ** returns 1 if there is a problem, 0 otherwise

+ **

+ ** Note that this function does not handle missing data (ie NA)

+ **

+ ********************************************************/

+

+

+void qnorm_c_handleNA(double *data, int *rows, int *cols){

+

+  double *target = Calloc(*rows,double);

+    

+  qnorm_c_determine_target(data, rows, cols, target, rows);

+  qnorm_c_using_target(data, rows, cols, target, rows);

+

+  Free(target);

+

+}

+

+

+/*********************************************************

+ **

+ ** SEXP R_qnorm_c_handleNA(SEXP X)

+ **

+ ** SEXP X      - a matrix

+ ** SEXP copy   - a flag if TRUE then make copy

+ **               before normalizing, if FALSE work in place

+ **               note that this can be dangerous since

+ **               it will change the original matrix.

+ **

+ ** returns a quantile normalized matrix.

+ **

+ ** This is a .Call() interface for quantile normalization

+ **

+ *********************************************************/

+

+SEXP R_qnorm_c_handleNA(SEXP X, SEXP copy){

+

+  SEXP Xcopy,dim1;

+  double *Xptr;

+  int rows,cols;

+  

+  PROTECT(dim1 = getAttrib(X,R_DimSymbol));

+  rows = INTEGER(dim1)[0];

+  cols = INTEGER(dim1)[1];

+  if (asInteger(copy)){

+    PROTECT(Xcopy = allocMatrix(REALSXP,rows,cols));

+    copyMatrix(Xcopy,X,0);

+  } else {

+    Xcopy = X;

+  }

+  Xptr = NUMERIC_POINTER(AS_NUMERIC(Xcopy));

+  

+  qnorm_c_handleNA(Xptr, &rows, &cols);

+  if (asInteger(copy)){

+    UNPROTECT(2);

+  } else {

+    UNPROTECT(1);

+  }

+  return Xcopy;

+}

+

+

+

 /*****************************************************************************************************

  *****************************************************************************************************

  **