[Bioc-devel] February NEWS

James MacDonald jmacdon at med.umich.edu
Tue Mar 6 19:05:25 CET 2007

	February 2007

	Modified to use ExpressionSet instead of exprSet

	Now readCelunits() can handle unit groups for which there are
	no probes, e.g., when stratifying on PM in a unit containing
	only MMs.

	Added writeCdfHeader(), writeCdfQcUnits() and writeCdfUnits(),
	all of which are used by writeCdf(). They make it possible to
	write a CDF in chunks using bounded memory.
	Updated AmiGO URL

	Changed structure of class BeadLevelList to handle uneven
	numbers of beads per array, and two-colour data. Updated
	plotting and summarising functions.
	Added boxplotBeads(), arrayNames(), numBeads(), getArrayData()
	to work on this class.

	New function readIllumina() to read in raw bead-level data.

	Added experimental unsaveSetSlot for internal use. This
	functiona llows one to set the slots of an S4 object in place
	without any copying. This should allow for considerable speed
	up when used appropriately.

	Added ExpressionSet introduction vignette. Moved old Biobase
	vignette to new 'legacy' directory.

	Disallowed drop = TRUE for AnnotatedDataFrame.


	Added port attribute to listMarts() and useMart() so users can
	choose which port to connect to when using biomaRt in MySQL
	mode. Changed default Ensembl host to martdb.ensembl.org using
	port 3316.
	Added retrieval of more sequence types with the getSequence
	function (such as retrieval of exons and upstream
	sequences). One can now also use filters for the type argument
	of getSequence(). Removed getXref() and getPossibleXrefs() as
	these have been deprecated by getBM() and getLDS(). Removed
	the martTable object as ouput from all functions is now a


	Added the Boyer-Moore algorithm. The defaults for
	matchPattern() are now fixed = TRUE and the 'auto' feature ow
	picks up the 'boyer-moore' algorithm whenever possible
	(generally faster than the shift-or algorithm.

	Added the 'naive' algorithm for QC and as a benchmark.

	Added the 'gregexpr' algorithm when pattern and subject are
	both standard character strings. The 'gregexpr' algorithm is a
	modified version of the standard R gregexpr(), to be used as a
	'pure R' benchmark.

	Changed modified gregexpr() to gregexpr2() and added
	gregexpr(). Both undocumented.

	Terminology changes. The 'Strong Good Suffix shifts' are now
	called the 'Very Strong Good Suffix shifts' to indicate the
	difference (>=) from the usual Strong Good Suffix shifts. The
	'Good Right Shifts' are now called the 'Matching Window

	Added ability to run huperGTest() using chromosome bands as
	categories. This adds the ChrMapHyperGParams and
	ChrMapHyperGResult classes and code.

	Improved handling of MAP annotation ranges. Instead of
	dropping annotations given as a range of chromosome bands, we
	now take the longest common prefix. There are still some
	inconsistencies in the annotation which we don't handle
	Removed unused geneCounts and universeCounts from HyperGResult
	object. Methods with the same names are available for
	accessing these data from the result object. Thanks to N. Le
	Meur for the bug report.


	Added two new summaries that can be used in
	summarizeReplicates(), 'closestToZero' and 'farthestFromZero'.

	Bug fix: memory of the size of unpacked image was lost on
	every operation using ImageMagick built in filters. Other
	filters were not affected, nor were IO, conversions or object
	detection. Everyone using the package should update to this

	Added propagate(), a Voronoi-based segmentation on image
	manifolds with modified metric.

	gene.details(), transcript.details() and exon.details() now
	take a vector as input and return a data.frame (as
	advertised). pc() now uses rowttests(), and does not need
	members to be specified if factor has exactly two levels.

	Added gene.strip() and probesets.in.eset().

	Improved scale on plot.gene().


	Changed optical background correction to tolerate no MM
	Added edgeMatrix method for graphAM.
	Fixed graphNEL ==> graphAM by removing rownames from
	adjMat. The optimized version was resulting in a graphAM
	instance with rownames and colnames on the adjMat. The graphAM
	initialize method removes rownames so the node names are
	stored once.

	Fixed bug in ftM2graphNEL() that was due to partial string
	matching in accessing list elements by name.


	normalizeRobustSpline() now works with only one print-tip
	group. New function mergeScahsRG().

	Deprecated getColClasses() and namesInFun() and replaced with
	read.columns(), a more general function that is similar to
	read.delim() but reads specified columns only. Argument '...'
	added to all as.matrix methods for compatibility with R-2.5.0.

	Bugfix: lm.series() and gls.series() were returning errors
	when the design matrix was not of full rank and the columns
	were names.

	Bugfix: write.fit() was using ambiguous column names for the
	coef, t-statistic and p-value when the MArrayLM object had
	only one column and these components were matrices with one
	named column.

	Default for digits changed in write.fit().


	Updated lumiR() to allow directly adding nuID when reading the
	data, and updated addNuId2lumi to add the tracking history.

	Updated the combine method of LumiBatch class, addNuId2lumi(),
	lumiR() for better adaptiveness.


	Added support for non-linear PCA and made corresponding
	additions to the documentation. Also added simulated example
	data for use with nlpca(). Changed Q2 cross-validation to
	guarantee that no full column or row is ever deleted. The
	screeplot() function was removed and replaced with plotR2().

	K-estimate now supports NMRMSEP and Q2 as error measure.

	Upgraded plotPcs() so it works with Sweave() and gives
	prettier output by basing it on pairs() instead of slplot().

	Added a new method kEstimateFast(). This also estimates the
	best number of components/similar genes for missing value
	estimation. In contrast to kEstimate() no crossvalidation is
	performed. This is a more rough estimate than provided by

	Improved speed of kEstimate(). Now possible to get the
	estimation error for each individual incomplete variable. This
	allows one to easily see for which variables missing value
	imputation makes sense.

	Removed ellipse dependency.

	Added vignette on missing value estimation.

	New method to comvert pcaRes to exprSet. pca() no longer
	returns an exprSet if an exprSet was used as input, use
	asExprSet() instead.


	Added C. elegans and yeast2 to data for QC stats. Updated to
	work with ExpressionSet class.


	Improvements on the predict method for vsn(). justvsn() is now
	Updated lymphoma dataset to ExpressionSet.
	Made vsn2() compatible with limma package.

	Updated vignette.

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues.

More information about the Bioc-devel mailing list