SC:19 Aug 2009 read.mir deliveres an RGList or EList depending on the presence of Bg values SC:20 Aug 2009 corrected bug in lvs.EList when supplying summarize.args added some S4 methods for EList object so to deal with EList or ExpressionSet the same way (v 0.1.2) SC:1 Sept 2009 added methods EList and RGList to estVC (v0.1.3) added methods for RGList to utility functions (see utils.R) SC:11 Sept 2009 corrected bug in summarize.EList SC:11 Sept 2009 added vsn normalization option SC:14 Sept 2009 vsn might work on G-Gb SC:19 Sept 2009 median polish now allows (unbalanced #) replication YP:4 Oct 2009 add options to estVC.std - weighted or asymptotic variance formula - joint mean and variance model, or only mean model - remove contr.helmert (revert to contr.treatment) - output inclues ArrayEffects as expression index SC:10 Oct 2009 fixed summarize.R and normalize.R functions to adapt to new estVC output - within "lvs", using summarize.args=list(method="rlm") would use Array Effects estimated by estVC. SC:14 Oct 2009 Summarization default now is "rlm" using Array Effects from RLM computation. SC:16 Oct 2009 added option (Atransf=c("sqrt","log")) to choose the transformation of Array Chi-squared (originally only sqrt). plot method modified accordingly. SC: 22 Oct 2009 Bug fixed in read.mir, as skip argument didn't account for automatic shift in connection pointer. Now use of seek (for OS where seek works) or reopening connection (Windows) would place the pointer back to the beginnig SC: 18 Nov 2009 v 0.2.4 Added option for stratified identification of lvs set, based on quantiles of reference distribution (default: median i.e. n.strata =2) SC: 20 Nov 2009 v 0.2.5 added option for parallel computation using either multicore or snow SC: 26 Nov 2009 v 0.2.5 method default changed to "joint" in estVC SC: 21 Jan 2010 v 0.2.6 maxit in gamma.glm is 20 and residuals are checked for being bigger than .Machine$double.eps ^ 0.5 instead of zero SC: 22 Jan 2010 v 0.2.7 - Allow to use affyPara for vsn computation. - Fixed bug on ArrayEffects ordering SC: 25 Jan 2010 v 0.2.8 - vsn parallel computation implemented without using affyPara in order to use "multicore". For relatively small samples using snow the way affyPara does increases the data transfer burden (we now split along cols) which make the program running slower. It might me useful for bigger sample size (how much?). Using multicore avoids data transfer to cores. SC: 26 Jan 2010 v 0.3.0 - C version for rlm and joint fit. SC: 25 Jan 2010 v 0.3.1 - Changed split to adapt to number of cores. This would affect only parallel computation SC: 02 Feb 2010 v 0.3.2 - Fixed some bugs in C code (max_iter was not set properly). - Removed splitting and in 0.3.1 as it doesn't improve computation time. - Removed parallel vsn as it must be better setup. TODO - Added option for plotting quantile line in plot.RA - Now selection on LVS probes can be done also setting an horizontal line - Now lvs accepts as argument a named vector that specify the LVS probes (selected outside) SC: 09 Feb 2010 v 0.3.3 - lvs allows to keep the LVS probe names as an attribute SC: 08 Mar 2010 v 0.3.4 - fixed bug in using level=probe SC: 22 Mar 2010 v 0.3.5 - little change in plot.RA (arg col.black and lwd=1.5 in rq line) SC: 24 Mar 2010 v 0.3.6 - Added converged and # of iterations in C code for rlm fit SuoChen: 24 May 2010 v 0.3.7 - Added man pages, examples and data SC: 25 May 2010 v 0.3.8 - Fixed man and some warnings in R CMD check SC: 04 Jun 2010 v 0.3.9 - Added a mixed VSN-smooth normalization method: normalization based on smooth-spline followed by variance stabilization without calibration (method="mixed") SC: 17 Jun 2010 v 0.3.10 - Added vignette SC: 30 Aug 2010 v 0.3.11 - Added NAMESPACE and modified "internal" functions to start with ".". Also some fix for parallel execution in Windows. SC: 30 Aug 2010 v 0.9.1 - First version for submission to Bioconductor SC: 15 Sep 2010 v 0.9.3 - First revision for submission to Bioconductor SC: 20 Sep 2010 v 0.9.4 - Second revision for submission to Bioconductor. Removed usage of . for hidden functions. Fixed vignette SC: 08 Dec 2010 v 1.1.2 - Fixed switch() in plot (utils.R) syntax (only 1 unnamed argument when the first is character) SC: 12 Apr 2014 v 1.3.4 - Fixed some warnings from R 3.1.0 - ream.mir now simply calls read.maimage from limma