Changes in version 1.26.0                        

    o   filterWindowsGlobal() finally behaves correctly for
	variable-width data=.

    o   normFactors() and normOffsets() accept DGEList inputs in their
	object= and se.out= arguments.

    o   mergeResults() and friends now default to taking tab= from the
	mcols() of the inputs.

                       Changes in version 1.24.0                        

    o   Accept a list of BamFiles in bam.files= for all functions.

                       Changes in version 1.22.0                        

    o   Removed deprecated functionality in consolidateTests() and
	related functions.

    o   Modified the method used to count the number of up/down windows
	in combineTests(). Reported a representative test in the output
	along with its log-fold changes.

    o   Renamed mixedClusters() to mixedTests() with deprecation of the
	former.

    o   Added minimalTests() for cluster-level summarization with a
	minimum proportion of significant tests.

    o   Standardized the output of getBestTest(), empiricalFDR(),
	mixedTests() and minimalTests() to that of combineTests(). In
	particular, getBestTest() now returns the index of its best
	window in the rep.test field.

                       Changes in version 1.20.0                        

    o   Removed deprecated functionality in normOffsets(), readParam(),
	scaledAverage().

    o   Added mergeResults(), overlapResults() wrapper functions to
	simplify getting region-level results.

    o   Added mergeWindowsList(), findOverlapsList() functions for
	consolidating windows from multiple analyses. Deprecated
	consolidateWindows().

    o   Added mergeResultsList(), overlapResultsList() wrapper
	functions to obtain consolidated region-level results.
	Deprecated consolidateTests(), consolidateOverlaps().

    o   Renamed regions= to ranges= in mergeWindows() for consistency.

    o   Added clusterWindowsList() to replace consolidateClusters().

    o   Added filterWindowsGlobal(), filterWindowsLocal(),
	filterWindowsProportion() and filterWindowsControl().
	Deprecated filterWindows().

    o   Minor renaming of scaleControlInfo() arguments, added
	assay.data= and assay.back= arguments.

                       Changes in version 1.18.0                        

    o   Deprecated the type= argument in normOffsets().

    o   Removed the deprecated consolidateSizes() function.

    o   Deprecated the use of BPPARAM= in readParam(). Added the
	BPPARAM= argument to all relevant functions.

    o   Modified csawUsersGuide() to point to the workflow package.

                       Changes in version 1.16.0                        

    o   Added normFactors() function to avoid confusion when
	normOffsets() returns factors.

    o   Deprecated type="scaling" option in normOffsets().

    o   Added calculateCPM() function for convenient calculation of
	(log-)CPMs.

    o   Split up consolidateSizes() function into consolidateWindows(),
	consolidateTests() and consolidateOverlaps(). Deprecated
	consolidateSizes() itself.

    o   Switched output of combineTests() and getBestTest() and related
	functions to a DataFrame.

    o   Modified mergeWindows() behaviour with specified sign=, for
	dealing with nested windows of opposing sign.

    o   Altered controlClusterFDR() to take the largest adjusted
	p-value threshold that yields a cluster-level FDR below
	target=.

    o   Simplified detailRanges() output so that it no longer returns
	an arbitrary exon number.

                       Changes in version 1.14.0                        

    o   Fully removed support for paramList objects.

    o   Removed support for normalize(), modified default option for
	se.out= in normOffsets().

                       Changes in version 1.12.0                        

    o   Removed support for paramList objects.

    o   Added option for normOffsets() to return SummarizedExperiment
	objects containing normalization data.

    o   Deprecated normalize() to avoid S4 method clashes.

    o   Moved scaling prior to control-based filtering into a new
	function, scaleControlFilter(), for greater modularity.

    o   Updated user's guide.

                       Changes in version 1.10.0                        

    o   
	Added calculation of dominant directionality in combineTests().
	Fixed out-of-array indexing bug in the C++ code.

    o   
	Supported factor input for ids argument in combineTests(),
	getBestTest().

    o   
	Added the empiricalFDR(), empiricalOverlaps() functions for
	controlling the empirical FDR.

    o   
	Added the mixedClusters(), mixedOverlaps() functions for
	testing for mixed clusters.

    o   
	Ensured that window-level FDR threshold chosen by
	controlClusterFDR() is not above the cluster-level FDR.

    o   
	Minor fix to scaledAverage() to avoid slightly inaccurate
	results. Also, zero or negative scale factors now return -Inf
	and NA, respectively.

    o   
	Switched to new scaleOffset() function for adding offsets in
	asDGEList(). Added option to specify the assay to be used.

    o   
	Added multi-TSS support in detailRanges().

    o   
	Modified paired-end machinery in windowCounts(), getPESizes()
	to be more accommodating of overruns.

    o   
	Ignored secondary and supplementary alignments in all
	functions.

    o   
	Added options to specify assay in SE objects in
	filterWindows().

    o   
	Replaced weighting with normalization options in
	profileSites().

    o   
	Updated user's guide.

                        Changes in version 1.8.0                        

    o   Added protection against NA values in filterWindows().

    o   Deprecated the use of parameter lists in any param= arguments.

    o   Tightened up allowable values of ext= arguments in various
	functions.

    o   Added the BPPARAM slot in the readParam class to store
	BiocParallelParam objects.

    o   Added support for parallelization in windowCounts(),
	regionCounts() and others.

    o   Updated documentation, user's guide.

                        Changes in version 1.6.0                        

    o   
	Restored normalize() as a S4 method returning a
	RangedSummarizedExperiment object.

    o   
	Modified asDGEList() to use any available normalization data in
	the input object.

    o   
	Generalized S4 methods to apply on SummarizedExperiment
	objects.

    o   
	Removed the rescue.ext option for PE handling, to maintain
	consistent totals calculations.

    o   
	Removed the fast.pe option for PE data handling, in favour of
	improved default processing.

    o   
	Removed dumpPE(), which is not required without the fast.pe
	option.

    o   
	Removed makeExtVector() in favour of list/DataFrame
	specification.

    o   
	windowCounts() and regionCounts() now compute and store the
	mean PE size and read length.

    o   
	Minor fix in correlateReads() for end-of-chromosome behaviour.

    o   
	Modified checkBimodality() so that the width argument behaves
	like ext in windowCounts().

    o   
	extractReads() with as.reads=TRUE for PE data now returns a
	GRangesList.

    o   
	Added the controlClusterFDR(), clusterWindows() and
	consolidateClusters() functions to automate control of the
	cluster-level FDR.

    o   
	Added protection against NA values in the cluster IDs for
	combineTests(), getBestTest(), upweightSummits().

    o   
	All read extraction methods are now CIGAR-aware and will ignore
	soft-clipped parts of the alignment.

                        Changes in version 1.4.0                        

    o   
	Added clusterFDR() function to compute the FDR for clusters of
	DB windows.

    o   
	Added checkBimodality() function to compute bimodality scores
	for regions.

    o   
	Modified normalize(), asDGEList() to allow manual specification
	of library sizes.

    o   
	Switched from normalizeCounts(), normalize() to S4 method
	normOffsets().

    o   
	Modified default parameter specification in strandedCounts(),
	to avoid errors.

    o   
	Switched to warning from error when a restricted chromosome is
	specified in extractReads().

    o   
	Modified extractReads() interface with improved support for
	extended read and paired read extraction.

    o   
	Added normalization options to filterWindows() when using
	control samples.

    o   
	Fixed bug for proportional filtering in filterWindows().

    o   
	Allowed correlateReads() to accept paired-end specification
	when extracting data.

    o   
	Added maximizeCcf() function to estimate the average fragment
	length.

    o   
	Added support for strand-specific overlapping in
	detailRanges().

    o   
	Increased the fidelity of retained information in dumped BAM
	file from dumpPE().

    o   
	Modified strand specification arguments for profileSites(),
	allowed reporting of individual profiles.

    o   
	Removed param= specification from wwhm().

    o   
	Switched to RangedSummarizedExperiment conventions for all
	relevant functions.

    o   
	Switched to mapqFilter for scanBam() when filtering on mapping
	quality.

    o   
	Added tests for previously untested functions.

    o   
	Slight updates to documentation, user's guide.

                        Changes in version 1.2.0                        

    o   
	Added getWidths(), scaledAverage() and filterWindows(), to
	facilitate comparison of abundances during filtering.

    o   
	Added findMaxima() to identify locally maximal windows from
	range data.

    o   
	Added profileSites() to examine the coverage profile around
	specified regions, with wwhm() to guess the ideal window size
	from the profile.

    o   
	Changed default window width in windowCounts() to 50 bp,
	default filter to a fixed count of 10. Also, filter=0 is
	honored when bin=TRUE.

    o   
	Switched from the depracated rowData to rowRanges for all
	manipulations of SummarizedExperiment.

    o   
	Changed all instances of `pet' to `pe' in read parameter
	specification, and renamed getPETSizes() to getPESizes().

    o   
	Removed the redundant rescue.pairs parameter in readParam().

    o   
	Added fast.pe parameter in readParam(), for fast paired-end
	data extraction. Added dumpPE() to pre-process paired-end BAM
	files for fast downstream extraction.

    o   
	Added support for custom column specification in getBestTest(),
	combineTests().

    o   
	Switched from reporting average log-FC to numbers of up/down
	windows in combineTests().

    o   
	Allowed getBestTest() to return all fields associated with the
	best window in the output table.

    o   
	Added upweightSummits() to compute weights favouring
	high-abundance windows.

    o   
	Added combineOverlaps(), getBestOverlaps() and summitOverlaps()
	wrapper functions for processing of Hits objects.

    o   
	Added consolidateSizes(), to consolidate DB results from
	multiple window sizes.

    o   
	Added support for custom key/name specification in
	detailRanges() for non-human/mouse systems.

    o   
	Added support for strand-specific read extraction in
	readParam(), strand-specific counting via strandedCounts().

    o   
	Added strand-awareness to mergeWindows(). Added protection
	against stranded input regions in extractReads(),
	detailRanges().

    o   
	Changed algorithm for splitting of large peaks in
	mergeWindows().

    o   
	Stored counting parameters in exptData for windowCounts(),
	regionCounts().

    o   
	Fixed small inaccuracies with continuity correction addition in
	normalizeCounts() for NB-loess.

    o   
	Switched to fragment midpoint for binning of paired-end data in
	windowCounts().

    o   
	Added support for lists of library-specific readParam objects
	in windowCounts(), regionCounts(), correlateReads().

    o   
	Added makeExtVector(), to support variable read extension
	lengths between libraries in windowCounts(), regionCounts().

    o   
	Added support for read extension within extractReads().

    o   
	Updated the user's guide to reflect new and modified functions.

    o   
	Added sra2bam.sh in inst/doc to reproducibly generate BAM files
	to run UG examples.

    o   
	Cleaned up code in inst/tests for modified functions, added new
	tests for new functions.

                        Changes in version 1.0.0                        

    o   
	New package csaw, for de novo detection of differential binding
	in ChIP-seq data.